deblan-mirror/buzz
1
0
Fork 0
mirror of https://github.com/chidiwilliams/buzz.git synced 2026-03-14 14:45:46 +01:00
Code Issues Projects Releases Packages Wiki Activity

Adding support for Vulkan for whisper.cpp (#1194)

Browse source
This commit is contained in:
Raivis Dejus 2025-07-20 16:53:36 +03:00 • committed by GitHub
commit 93afbead37
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
50 changed files with 1161 additions and 382 deletions
Download patch file Download diff file Expand all files Collapse all files

3
.coveragerc
View file

@ -1,6 +1,7 @@
[run]
omit =
buzz/whisper_cpp.py
buzz/whisper_cpp/*
buzz/whisper_cpp_vulkan/*
*_test.py
demucs/*

37
.github/workflows/ci.yml vendored
View file

@ -70,8 +70,9 @@ jobs:
poetry add torch==2.2.2 torchaudio==2.2.2
if: "matrix.os == 'macos-13'"
- name: Install dependencies
run: poetry install
- name: Add msbuild to PATH
uses: microsoft/setup-msbuild@v2
if: runner.os == 'Windows'
- name: Install apt dependencies
run: |
@ -82,6 +83,15 @@ jobs:
sudo apt-get install libyaml-dev libxkbcommon-x11-0 libxcb-icccm4 libxcb-image0 libxcb-keysyms1 libxcb-randr0 libxcb-render-util0 libxcb-xinerama0 libxcb-shape0 libxcb-cursor0 libportaudio2 gettext libpulse0 libgl1-mesa-dev
if: "startsWith(matrix.os, 'ubuntu-')"
- name: Install Vulkan SDK
uses: humbletim/install-vulkan-sdk@v1.2
with:
version: 1.4.309.0
cache: true
- name: Install dependencies
run: poetry install
- name: Test
run: |
poetry run make test
@ -132,6 +142,12 @@ jobs:
- uses: AnimMouse/setup-ffmpeg@v1.2.1
id: setup-ffmpeg
- name: Install Vulkan SDK
uses: humbletim/install-vulkan-sdk@v1.2
with:
version: 1.4.309.0
cache: true
# Adding windows Nvidia CUDA torch and its dependencies
# Linux only packages were removed from the list.
- name: Install CUDA torch for Windows
@ -153,6 +169,10 @@ jobs:
poetry add torch==2.2.2 torchaudio==2.2.2 ctranslate2==4.3.1
if: "matrix.os == 'macos-13'"
- name: Add msbuild to PATH
uses: microsoft/setup-msbuild@v2
if: runner.os == 'Windows'
- name: Install dependencies
run: poetry install
@ -255,19 +275,12 @@ jobs:
with:
submodules: recursive
- name: Copy Windows DLLs
run: |
if [ "$RUNNER_OS" == "Windows" ]; then
cp -r ./dll_backup ./buzz/
fi
shell: bash
- name: Build wheels
uses: pypa/cibuildwheel@v2.22.0
env:
CIBW_ARCHS_WINDOWS: "auto"
CIBW_ARCHS_MACOS: "universal2"
CIBW_ARCHS_LINUX: "auto"
CIBW_ARCHS_WINDOWS: "auto64"
CIBW_ARCHS_MACOS: "auto64"
CIBW_ARCHS_LINUX: "auto64"
- uses: actions/upload-artifact@v4
with:

2
.github/workflows/snapcraft.yml vendored
View file

@ -21,7 +21,7 @@ jobs:
- name: Maximize build space
uses: easimon/maximize-build-space@master
with:
root-reserve-mb: 20000
root-reserve-mb: 25000
swap-size-mb: 1024
remove-dotnet: 'true'
remove-android: 'true'

12
.gitignore vendored
View file

@ -13,13 +13,13 @@ coverage.xml
venv/
# whisper_cpp
libwhisper.*
libwhisper-coreml.*
whisper_cpp
whisper_cpp.exe
whisper.dll
buzz/whisper_cpp.py
buzz/whisper_cpp_coreml.py
*.exe
*.dll
*.dylib
*.so
buzz/whisper_cpp/*
buzz/whisper_cpp_vulkan/*
# Internationalization - compiled binaries
*.mo

51
Buzz.spec
View file

@ -1,4 +1,5 @@
# -*- mode: python ; coding: utf-8 -*-
import os
import os.path
import platform
import shutil
@ -43,26 +44,50 @@ if DEBUG:
else:
options = []
binaries = [
(
"buzz/whisper.dll" if platform.system() == "Windows" else "buzz/libwhisper.*",
".",
),
(shutil.which("ffmpeg"), "."),
(shutil.which("ffprobe"), "."),
]
def find_dependency(name: str) -> str:
paths = os.environ["PATH"].split(os.pathsep)
candidates = []
for path in paths:
exe_path = os.path.join(path, name)
if os.path.isfile(exe_path):
candidates.append(exe_path)
# Include libwhisper-coreml.dylib on Apple Silicon
if platform.system() == "Darwin" and platform.machine() == "arm64":
binaries.append(("buzz/libwhisper-coreml.dylib", "."))
# Check for chocolatery shims
shim_path = os.path.normpath(os.path.join(path, "..", "lib", "ffmpeg", "tools", "ffmpeg", "bin", name))
if os.path.isfile(shim_path):
candidates.append(shim_path)
if not candidates:
return None
# Pick the largest file
return max(candidates, key=lambda f: os.path.getsize(f))
if platform.system() == "Windows":
binaries = [
(find_dependency("ffmpeg.exe"), "."),
(find_dependency("ffprobe.exe"), "."),
]
else:
binaries = [
(shutil.which("ffmpeg"), "."),
(shutil.which("ffprobe"), "."),
]
if platform.system() == "Linux":
binaries.append(("buzz/whisper_cpp/*.so", "buzz/whisper_cpp"))
binaries.append(("buzz/whisper_cpp_vulkan/*.so", "buzz/whisper_cpp_vulkan"))
if platform.system() == "Darwin":
binaries.append(("buzz/whisper_cpp/*.dylib", "buzz/whisper_cpp"))
# Include dll_backup folder and its contents on Windows
if platform.system() == "Windows":
datas += [("dll_backup", "dll_backup")]
datas += collect_data_files("msvc-runtime")
binaries.append(("dll_backup/SDL2.dll", "dll_backup"))
binaries.append(("dll_backup/whisper.dll", "dll_backup"))
binaries.append(("buzz/whisper_cpp/*.dll", "buzz/whisper_cpp"))
binaries.append(("buzz/*.exe", "."))
a = Analysis(
["main.py"],

149
Makefile
View file

@ -12,35 +12,23 @@ bundle_mac: dist/Buzz.app codesign_all_mac zip_mac notarize_zip staple_app_mac d
bundle_mac_unsigned: dist/Buzz.app zip_mac dmg_mac_unsigned
UNAME_S := $(shell uname -s)
UNAME_M := $(shell uname -m)
LIBWHISPER :=
ifeq ($(OS), Windows_NT)
LIBWHISPER=whisper.dll
else
ifeq ($(UNAME_S), Darwin)
LIBWHISPER=libwhisper.dylib
else
LIBWHISPER=libwhisper.so
endif
endif
clean:
ifeq ($(OS), Windows_NT)
-del /f buzz\$(LIBWHISPER) 2> nul
-del /f buzz\whisper_cpp.py 2> nul
-rmdir /s /q whisper.cpp\build 2> nul
-rmdir /s /q dist 2> nul
-rm -f buzz/$(LIBWHISPER)
-rm -f buzz/whisper_cpp.py
-rm -rf whisper.cpp/build || true
-rm -rf dist/* || true
-rmdir /s /q buzz\whisper_cpp
-rmdir /s /q buzz\whisper-server.exe
-rmdir /s /q whisper.cpp\build
-rmdir /s /q dist
-Remove-Item -Recurse -Force buzz\whisper_cpp
-Remove-Item -Recurse -Force buzz\whisper-server.exe
-Remove-Item -Recurse -Force whisper.cpp\build
-Remove-Item -Recurse -Force dist\*
-rm -rf buzz/whisper_cpp
-rm -fr buzz/whisper-server.exe
-rm -rf whisper.cpp/build
-rm -rf dist/*
else
rm -f buzz/$(LIBWHISPER)
rm -f buzz/whisper_cpp.py
rm -f buzz/libwhisper-coreml.dylib || true
rm -f buzz/whisper_cpp_coreml.py || true
rm -rf buzz/whisper_cpp || true
rm -fr buzz/whisper_cpp_vulkan || true
rm -rf whisper.cpp/build || true
rm -rf dist/* || true
endif
@ -60,53 +48,74 @@ version:
poetry version ${version}
echo "VERSION = \"${version}\"" > buzz/__version__.py
CMAKE_FLAGS=
ifeq ($(UNAME_S),Darwin)
AVX1_M := $(shell sysctl machdep.cpu.features)
ifeq (,$(findstring AVX1.0,$(AVX1_M)))
CMAKE_FLAGS += -DWHISPER_NO_AVX=ON
endif
ifeq (,$(findstring FMA,$(AVX1_M)))
CMAKE_FLAGS += -DWHISPER_NO_FMA=ON
endif
AVX2_M := $(shell sysctl machdep.cpu.leaf7_features)
ifeq (,$(findstring AVX2,$(AVX2_M)))
CMAKE_FLAGS += -DWHISPER_NO_AVX2=ON
endif
CMAKE_FLAGS += -DCMAKE_OSX_ARCHITECTURES="x86_64;arm64"
else
ifeq ($(OS), Windows_NT)
CMAKE_FLAGS += -DBUILD_SHARED_LIBS=ON -DCMAKE_BUILD_TYPE=Release
endif
endif
buzz/$(LIBWHISPER):
buzz/whisper_cpp:
ifeq ($(OS), Windows_NT)
cp dll_backup/whisper.dll buzz || copy dll_backup\whisper.dll buzz\whisper.dll
cp dll_backup/SDL2.dll buzz || copy dll_backup\SDL2.dll buzz\SDL2.dll
else
cmake -S whisper.cpp -B whisper.cpp/build/ $(CMAKE_FLAGS)
cmake --build whisper.cpp/build --verbose
cp whisper.cpp/build/bin/Debug/$(LIBWHISPER) buzz || true
cp whisper.cpp/build/$(LIBWHISPER) buzz || true
endif
# Build CoreML support on ARM Macs
ifeq ($(shell uname -m), arm64)
ifeq ($(shell uname -s), Darwin)
rm -rf whisper.cpp/build || true
cmake -S whisper.cpp -B whisper.cpp/build/ $(CMAKE_FLAGS) -DWHISPER_COREML=1
cmake --build whisper.cpp/build --verbose
cp whisper.cpp/build/bin/Debug/$(LIBWHISPER) buzz/libwhisper-coreml.dylib || true
cp whisper.cpp/build/$(LIBWHISPER) buzz/libwhisper-coreml.dylib || true
endif
# Build Whisper for CPU
# The _DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR is needed to prevent mutex lock issues on Windows
# https://github.com/actions/runner-images/issues/10004#issuecomment-2156109231
# -DCMAKE_[C|CXX]_COMPILER_WORKS=TRUE is used to prevent issue in building test program that fails on CI
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DCMAKE_INSTALL_RPATH='$$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON -DCMAKE_C_FLAGS="-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR" -DCMAKE_CXX_FLAGS="-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR" -DCMAKE_C_COMPILER_WORKS=TRUE -DCMAKE_CXX_COMPILER_WORKS=TRUE
cmake --build whisper.cpp/build -j --config Release --verbose
-mkdir buzz/whisper_cpp
cp dll_backup/SDL2.dll buzz/whisper_cpp
cp whisper.cpp/build/bin/Release/whisper.dll buzz/whisper_cpp
cp whisper.cpp/build/bin/Release/ggml.dll buzz/whisper_cpp
cp whisper.cpp/build/bin/Release/ggml-base.dll buzz/whisper_cpp
cp whisper.cpp/build/bin/Release/ggml-cpu.dll buzz/whisper_cpp
# Build Whisper with Vulkan support. On Windows whisper-server.exe wil lbe used as dll approach is unreliable,
# it often does not see the GPU
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=OFF -DCMAKE_INSTALL_RPATH='$$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON -DCMAKE_C_FLAGS="-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR" -DCMAKE_CXX_FLAGS="-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR" -DCMAKE_C_COMPILER_WORKS=TRUE -DCMAKE_CXX_COMPILER_WORKS=TRUE -DGGML_VULKAN=1
cmake --build whisper.cpp/build -j --config Release --verbose
cp whisper.cpp/build/bin/Release/whisper-server.exe buzz/
endif
buzz/whisper_cpp.py: buzz/$(LIBWHISPER) translation_mo
cd buzz && ctypesgen ../whisper.cpp/whisper.h -lwhisper -o whisper_cpp.py
ifeq ($(shell uname -m), arm64)
ifeq ($(shell uname -s), Darwin)
cd buzz && ctypesgen ../whisper.cpp/whisper.h -lwhisper-coreml -o whisper_cpp_coreml.py
ifeq ($(shell uname -s), Linux)
# Build Whisper for CPU
-rm -rf whisper.cpp/build || true
-mkdir -p buzz/whisper_cpp
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DCMAKE_INSTALL_RPATH='$$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON
cmake --build whisper.cpp/build -j --config Release --verbose
cp whisper.cpp/build/src/libwhisper.so buzz/whisper_cpp/libwhisper.so || true
cp whisper.cpp/build/ggml/src/libggml.so buzz/whisper_cpp || true
cp whisper.cpp/build/ggml/src/libggml-base.so buzz/whisper_cpp || true
cp whisper.cpp/build/ggml/src/libggml-cpu.so buzz/whisper_cpp || true
# Build Whisper for Vulkan
rm -rf whisper.cpp/build || true
-mkdir -p buzz/whisper_cpp_vulkan
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DCMAKE_INSTALL_RPATH='$$ORIGIN' -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON -DGGML_VULKAN=1
cmake --build whisper.cpp/build -j --config Release --verbose
cp whisper.cpp/build/src/libwhisper.so buzz/whisper_cpp_vulkan/whisper-vulkan.so || true
cp whisper.cpp/build/ggml/src/libggml.so buzz/whisper_cpp_vulkan || true
cp whisper.cpp/build/ggml/src/libggml-base.so buzz/whisper_cpp_vulkan || true
cp whisper.cpp/build/ggml/src/libggml-cpu.so buzz/whisper_cpp_vulkan || true
cp whisper.cpp/build/ggml/src/ggml-vulkan/libggml-vulkan.so buzz/whisper_cpp_vulkan || true
endif
# Build on Macs
ifeq ($(shell uname -s), Darwin)
-rm -rf whisper.cpp/build || true
-mkdir -p buzz/whisper_cpp
ifeq ($(shell uname -m), arm64)
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON -DWHISPER_COREML=1
else
cmake -S whisper.cpp -B whisper.cpp/build/ -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=ON
endif
cmake --build whisper.cpp/build -j --config Release --verbose
cp whisper.cpp/build/src/libwhisper.dylib buzz/whisper_cpp/ || true
cp whisper.cpp/build/ggml/src/libggml* buzz/whisper_cpp/ || true
endif
buzz/whisper_cpp.py: buzz/whisper_cpp translation_mo
cd buzz && ctypesgen ../whisper.cpp/include/whisper.h -I../whisper.cpp/ggml/include -lwhisper -o ./whisper_cpp/whisper_cpp.py
ifeq ($(shell uname -s), Linux)
cd buzz && ctypesgen ../whisper.cpp/include/whisper.h -I../whisper.cpp/ggml/include -lwhisper-vulkan -o ./whisper_cpp_vulkan/whisper_cpp_vulkan.py
endif
# Prints all the Mac developer identities used for code signing
@ -238,7 +247,7 @@ ifeq ($(OS), Windows_NT)
done
else
for dir in buzz/locale/*/ ; do \
python msgfmt.py -o $$dir/LC_MESSAGES/buzz.mo $$dir/LC_MESSAGES/buzz.po; \
python3 msgfmt.py -o $$dir/LC_MESSAGES/buzz.mo $$dir/LC_MESSAGES/buzz.po; \
done
endif

5
buzz/buzz.py
View file

@ -26,7 +26,10 @@ os.environ["PATH"] += os.pathsep + APP_BASE_DIR
# Add the app directory to the DLL list: https://stackoverflow.com/a/64303856
if platform.system() == "Windows":
os.add_dll_directory(APP_BASE_DIR)
os.add_dll_directory(os.path.join(APP_BASE_DIR, "dll_backup"))
dll_backup_dir = os.path.join(APP_BASE_DIR, "dll_backup")
if os.path.isdir(dll_backup_dir):
os.add_dll_directory(dll_backup_dir)
def main():

12
buzz/file_transcriber_queue_worker.py
View file

@ -37,11 +37,16 @@ class FileTranscriberQueueWorker(QObject):
super().__init__(parent)
self.tasks_queue = queue.Queue()
self.canceled_tasks: Set[UUID] = set()
self.current_transcriber = None
@pyqtSlot()
def run(self):
logging.debug("Waiting for next transcription task")
# Clean up of previous run.
if self.current_transcriber is not None:
self.current_transcriber.stop()
# Get next non-canceled task from queue
while True:
self.current_task: Optional[FileTranscriptionTask] = self.tasks_queue.get()
@ -131,9 +136,14 @@ class FileTranscriberQueueWorker(QObject):
def cancel_task(self, task_id: UUID):
self.canceled_tasks.add(task_id)
if self.current_task.uid == task_id:
if self.current_task is not None and self.current_task.uid == task_id:
if self.current_transcriber is not None:
self.current_transcriber.stop()
if self.current_transcriber_thread is not None:
if not self.current_transcriber_thread.wait(3000):
logging.warning("Transcriber thread did not terminate gracefully")
self.current_transcriber_thread.terminate()
def on_task_error(self, error: str):
if (

35
buzz/locale/ca_ES/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: buzz\n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-05-17 18:25+0200\n"
"Last-Translator: Éric Duarte <contacto@ericdq.com>\n"
"Language-Team: Catalan <jmas@softcatala.org>\n"
@ -295,8 +295,8 @@ msgid "Download failed"
msgstr "Descàrrega fallida"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Error"
@ -486,11 +486,11 @@ msgstr "Esperant la traducció de la IA..."
msgid "Microphone:"
msgstr "Micròfon:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "S'ha produït un error en iniciar un enregistrament nou:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -540,7 +540,6 @@ msgid "Export"
msgstr "Exporta"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Traduir"
@ -664,6 +663,24 @@ msgstr "Selecciona un fitxer d'àudio"
msgid "Unable to save OpenAI API key to keyring"
msgstr "No s'ha pogut desar la clau OpenAI API a l'anell de claus"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Configuració de la traducció"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Transcriure"
@ -1024,7 +1041,11 @@ msgstr "Sundanès"
msgid "Cantonese"
msgstr "Cantonès"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "S'ha produït un error de connexió"

35
buzz/locale/da_DK/LC_MESSAGES/buzz.po
View file

@ -2,7 +2,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: \n"
"Last-Translator: Ole Guldberg2 <xalt7x.service@gmail.com>\n"
"Language-Team: \n"
@ -295,8 +295,8 @@ msgid "Download failed"
msgstr "Download mislykkedes"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Fejl"
@ -487,11 +487,11 @@ msgstr "Venter på AI oversættelse..."
msgid "Microphone:"
msgstr "Mikrofon:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Der skete en fejl ved opstart af en ny optagelse:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -541,7 +541,6 @@ msgid "Export"
msgstr "Eksporter"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Oversæt"
@ -665,6 +664,24 @@ msgstr "Vælg audio-fil"
msgid "Unable to save OpenAI API key to keyring"
msgstr "Kan ikke gemme OpenAI API-nøgle i nøgleringen"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Oversættelsesindstillinger"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Transkriber"
@ -1025,7 +1042,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Der er opstået en forbindelsesfejl"

35
buzz/locale/de_DE/LC_MESSAGES/buzz.po
View file

@ -6,7 +6,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-03-05 14:41+0100\n"
"Last-Translator: \n"
"Language-Team: \n"
@ -295,8 +295,8 @@ msgid "Download failed"
msgstr "Der Download ist fehlgeschlagen"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Fehler"
@ -487,11 +487,11 @@ msgstr "Warten auf KI-Übersetzung..."
msgid "Microphone:"
msgstr "Mikrofon:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Beim Starten einer neuen Aufnahme ist ein Fehler aufgetreten:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -541,7 +541,6 @@ msgid "Export"
msgstr "Export"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Übersetzen"
@ -666,6 +665,24 @@ msgid "Unable to save OpenAI API key to keyring"
msgstr ""
"Der OpenAI-API-Schlüssel kann nicht im Schlüsselbund gespeichert werden"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Übersetzungseinstellungen"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Transkribieren"
@ -1026,7 +1043,11 @@ msgstr "Sundanesisch"
msgid "Cantonese"
msgstr "Kantonesisch"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Ein Verbindungsfehler ist aufgetreten"

34
buzz/locale/en_US/LC_MESSAGES/buzz.po
View file

@ -8,7 +8,7 @@ msgid ""
msgstr ""
"Project-Id-Version: PACKAGE VERSION\n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: YEAR-MO-DA HO:MI+ZONE\n"
"Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
"Language-Team: LANGUAGE <LL@li.org>\n"
@ -287,8 +287,8 @@ msgid "Download failed"
msgstr ""
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr ""
@ -477,11 +477,11 @@ msgstr ""
msgid "Microphone:"
msgstr ""
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr ""
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -529,7 +529,6 @@ msgid "Export"
msgstr ""
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr ""
@ -647,6 +646,23 @@ msgstr ""
msgid "Unable to save OpenAI API key to keyring"
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
msgid "Translate to English"
msgstr ""
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr ""
@ -1007,7 +1023,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr ""

35
buzz/locale/es_ES/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-05-17 18:27+0200\n"
"Last-Translator: Éric Duarte <contacto@ericdq.com>\n"
"Language-Team: \n"
@ -301,8 +301,8 @@ msgid "Download failed"
msgstr "Descarga fallida"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Error"
@ -517,12 +517,12 @@ msgid "Microphone:"
msgstr "Micrófono:"
# automatic translation
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Se produjo un error al iniciar una grabación nueva:"
# automatic translation
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -575,7 +575,6 @@ msgid "Export"
msgstr "Exportar"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Traducir"
@ -707,6 +706,24 @@ msgstr "Seleccionar archivo de audio"
msgid "Unable to save OpenAI API key to keyring"
msgstr "No se puede guardar la clave de la API de OpenAI en el llavero"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Ajustes de traducción"
# automatic translation
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
@ -1068,7 +1085,11 @@ msgstr "Sundanés"
msgid "Cantonese"
msgstr "Cantonés"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Se ha producido un error de conexión"

35
buzz/locale/it_IT/LC_MESSAGES/buzz.po
View file

@ -6,7 +6,7 @@ msgid ""
msgstr ""
"Project-Id-Version: buzz\n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-05-30 15:22+0100\n"
"Language-Team: (Italiano) Albano Battistella <albanobattistella@gmail.com>\n"
"Language: it_IT\n"
@ -294,8 +294,8 @@ msgid "Download failed"
msgstr "Download non riuscito"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Errore"
@ -486,11 +486,11 @@ msgstr "In attesa della traduzione AI..."
msgid "Microphone:"
msgstr "Microfono:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Si è verificato un errore durante l'avvio della nuova registrazione:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -540,7 +540,6 @@ msgid "Export"
msgstr "Esporta"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Tradurre"
@ -664,6 +663,24 @@ msgstr "Seleziona file audio"
msgid "Unable to save OpenAI API key to keyring"
msgstr "Impossibile salvare la chiave API OpenAI nel portachiavi"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Impostazioni di traduzione"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Trascrivere"
@ -1024,7 +1041,11 @@ msgstr "Sundanese"
msgid "Cantonese"
msgstr "Cantonese"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Si è verificato un errore di connessione"

35
buzz/locale/ja_JP/LC_MESSAGES/buzz.po
View file

@ -2,7 +2,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: \n"
"Last-Translator: nunawa <71294849+nunawa@users.noreply.github.com>\n"
"Language-Team: \n"
@ -291,8 +291,8 @@ msgid "Download failed"
msgstr "ダウンロード失敗"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "エラー"
@ -483,11 +483,11 @@ msgstr "AI翻訳を待っています..."
msgid "Microphone:"
msgstr "マイク:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "新規録音開始時にエラーが発生しました:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -537,7 +537,6 @@ msgid "Export"
msgstr "出力"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "翻訳"
@ -660,6 +659,24 @@ msgstr "音声ファイルを選択"
msgid "Unable to save OpenAI API key to keyring"
msgstr "OpenAI API キーをkeyringに保存できません"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "翻訳設定"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "文字起こし"
@ -1020,7 +1037,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "接続エラーが発生しました"

43
buzz/locale/lv_LV/LC_MESSAGES/buzz.po
View file

@ -8,8 +8,8 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"PO-Revision-Date: 2025-07-06 20:11+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-07-18 09:14+0300\n"
"Last-Translator: \n"
"Language-Team: \n"
"Language: lv_LV\n"
@ -297,8 +297,8 @@ msgid "Download failed"
msgstr "Lejupielāde neizdevās"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Kļūda"
@ -383,6 +383,8 @@ msgid ""
"Enter instructions for AI on how to translate, for example 'Please translate "
"each text sent to you from English to Spanish.'"
msgstr ""
"Ievadiet instrukcijas mākslīgajam intelektam, piemēram, 'Lūdzu, iztulko "
"katru tev atsūtīto tekstu no angļu valodas latviski'"
#: buzz/widgets/transcriber/advanced_settings_dialog.py:92
msgid "Instructions for AI:"
@ -489,11 +491,11 @@ msgstr "Gaida MI tulkojumu..."
msgid "Microphone:"
msgstr "Mikrofons:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Sākot jaunu ierakstu notikusi kļūda:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -543,7 +545,6 @@ msgid "Export"
msgstr "Eksportēt"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Tulkot"
@ -665,6 +666,28 @@ msgstr "Izvēlieties audio failu"
msgid "Unable to save OpenAI API key to keyring"
msgstr "Neizdevās saglabāt OpenAI API atslēgu atslēgu saišķī"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
"Whisper serverim neizdevās ieslēgties. Lūdzu pārbaudiet lietotnes žurnāla "
"ierakstus."
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
"Whisper serverim neizdevās ieslēgties, jo nepietika atmiņas. Lūdzu mēģiniet "
"vēlreiz ar mazāku modeli. Lai izmantotu tikai CPU iestatiet "
"BUZZ_FORCE_CPU=TRUE vides mainīgo."
#: buzz/transcriber/transcriber.py:24
msgid "Translate to English"
msgstr "Tulkot angliski"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Atpazīt"
@ -1025,7 +1048,11 @@ msgstr "Sundāņu"
msgid "Cantonese"
msgstr "Kantonas"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr "Palaiž Whisper.cpp..."
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Notika savienojuma kļūda"

35
buzz/locale/nl/LC_MESSAGES/buzz.po
View file

@ -8,7 +8,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-03-20 18:30+0100\n"
"Last-Translator: Heimen Stoffels <vistausss@fastmail.com>\n"
"Language-Team: none\n"
@ -297,8 +297,8 @@ msgid "Download failed"
msgstr "Het downloaden is mislukt"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Foutmelding"
@ -489,11 +489,11 @@ msgstr "Bezig met wachten op AI-vertaling…"
msgid "Microphone:"
msgstr "Microfoon:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Er is een fout opgetreden tijdens het starten van de opname:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -541,7 +541,6 @@ msgid "Export"
msgstr "Exporteren"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Vertalen"
@ -664,6 +663,24 @@ msgstr "Kies een audiobestand"
msgid "Unable to save OpenAI API key to keyring"
msgstr "De OpenAI-api-sleutel kan niet worden bewaard in de sleutelbos"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Vertaalinstellingen"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Transcriberen"
@ -1024,7 +1041,11 @@ msgstr "Soedanees"
msgid "Cantonese"
msgstr "Kantonees"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Er is een verbindingsfout opgetreden"

35
buzz/locale/pl_PL/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2024-03-17 20:50+0200\n"
"Last-Translator: \n"
"Language-Team: \n"
@ -298,8 +298,8 @@ msgid "Download failed"
msgstr "Pobrany"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Błąd"
@ -495,11 +495,11 @@ msgstr ""
msgid "Microphone:"
msgstr "Mikrofon:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Wystąpił błąd podczas rozpoczęcia nowego nagrania:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -550,7 +550,6 @@ msgid "Export"
msgstr ""
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr ""
@ -674,6 +673,24 @@ msgstr "Wybierz plik audio"
msgid "Unable to save OpenAI API key to keyring"
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Nowa transkrypcja"
#: buzz/transcriber/transcriber.py:25
#, fuzzy
msgid "Transcribe"
@ -1035,7 +1052,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr ""

35
buzz/locale/pt_BR/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: Buzz\n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2025-06-29 22:23-0300\n"
"Last-Translator: Paulo Schopf <pschopf@gmail.com>\n"
"Language-Team: none\n"
@ -295,8 +295,8 @@ msgid "Download failed"
msgstr "Falha ao baixar"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Erro"
@ -485,11 +485,11 @@ msgstr "Aguardando tradução da IA..."
msgid "Microphone:"
msgstr "Microfone:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "Ocorreu um erro ao iniciar uma nova gravação:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -539,7 +539,6 @@ msgid "Export"
msgstr "Exportar"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Traduzir"
@ -663,6 +662,24 @@ msgstr "Selecionar arquivo de áudio"
msgid "Unable to save OpenAI API key to keyring"
msgstr "Não foi possível salvar a chave da API OpenAI no cofre de chaves"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Configurações de tradução"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Transcrever"
@ -1023,7 +1040,11 @@ msgstr "Sundanês"
msgid "Cantonese"
msgstr "Cantonês"
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Ocorreu um erro de conexão"

35
buzz/locale/uk_UA/LC_MESSAGES/buzz.po
View file

@ -2,7 +2,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: \n"
"Last-Translator: Yevhen Popok <xalt7x.service@gmail.com>\n"
"Language-Team: \n"
@ -293,8 +293,8 @@ msgid "Download failed"
msgstr "Невдале завантаження"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "Помилка"
@ -485,11 +485,11 @@ msgstr "Очікування перекладу від ШІ..."
msgid "Microphone:"
msgstr "Мікрофон:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "При старті нового запису виникла помилка:"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -539,7 +539,6 @@ msgid "Export"
msgstr "Експорт"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "Перекласти"
@ -661,6 +660,24 @@ msgstr "Вибрати аудіофайл"
msgid "Unable to save OpenAI API key to keyring"
msgstr "Не вдається додати до звʼязки ключів API-ключ OpenAI"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "Налаштування перекладу"
#: buzz/transcriber/transcriber.py:25
msgid "Transcribe"
msgstr "Розпізнати"
@ -1021,7 +1038,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "Виникла помилка зʼєднання"

35
buzz/locale/zh_CN/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2023-05-01 15:45+0800\n"
"Last-Translator: \n"
"Language-Team: lamb \n"
@ -301,8 +301,8 @@ msgid "Download failed"
msgstr "下载模型失败"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr "错误"
@ -498,11 +498,11 @@ msgstr "等待AI翻译..."
msgid "Microphone:"
msgstr "麦克风:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "开始新录制时出错"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -551,7 +551,6 @@ msgid "Export"
msgstr "导出"
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr "翻译"
@ -674,6 +673,24 @@ msgstr "选择音频文件"
msgid "Unable to save OpenAI API key to keyring"
msgstr "无法将OpenAI API密钥保存到密钥串"
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "翻译设置"
#: buzz/transcriber/transcriber.py:25
#, fuzzy
msgid "Transcribe"
@ -1035,7 +1052,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr "连接发生错误"

35
buzz/locale/zh_TW/LC_MESSAGES/buzz.po
View file

@ -7,7 +7,7 @@ msgid ""
msgstr ""
"Project-Id-Version: \n"
"Report-Msgid-Bugs-To: \n"
"POT-Creation-Date: 2025-07-06 20:16+0300\n"
"POT-Creation-Date: 2025-07-18 09:13+0300\n"
"PO-Revision-Date: 2023-05-01 15:45+0800\n"
"Last-Translator: \n"
"Language-Team: Lamb\n"
@ -296,8 +296,8 @@ msgid "Download failed"
msgstr "下載模型"
#: buzz/widgets/preferences_dialog/models_preferences_widget.py:273
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:515
#: buzz/model_loader.py:529
#: buzz/widgets/main_window.py:295 buzz/model_loader.py:505
#: buzz/model_loader.py:519
msgid "Error"
msgstr ""
@ -493,11 +493,11 @@ msgstr ""
msgid "Microphone:"
msgstr "麥克風:"
#: buzz/widgets/recording_transcriber_widget.py:577
#: buzz/widgets/recording_transcriber_widget.py:579
msgid "An error occurred while starting a new recording:"
msgstr "開始新錄製出錯"
#: buzz/widgets/recording_transcriber_widget.py:581
#: buzz/widgets/recording_transcriber_widget.py:583
msgid ""
"Please check your audio devices or check the application logs for more "
"information."
@ -546,7 +546,6 @@ msgid "Export"
msgstr ""
#: buzz/widgets/transcription_viewer/transcription_viewer_widget.py:174
#: buzz/transcriber/transcriber.py:24
msgid "Translate"
msgstr ""
@ -668,6 +667,24 @@ msgstr "選擇聲音檔案"
msgid "Unable to save OpenAI API key to keyring"
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:51
#: buzz/transcriber/recording_transcriber.py:372
msgid "Whisper server failed to start. Check logs for details."
msgstr ""
#: buzz/transcriber/local_whisper_cpp_server_transcriber.py:54
#: buzz/transcriber/recording_transcriber.py:375
msgid ""
"Whisper server failed to start due to insufficient memory. Please try again "
"with a smaller model. To force CPU mode use BUZZ_FORCE_CPU=TRUE environment "
"variable."
msgstr ""
#: buzz/transcriber/transcriber.py:24
#, fuzzy
msgid "Translate to English"
msgstr "新錄製"
#: buzz/transcriber/transcriber.py:25
#, fuzzy
msgid "Transcribe"
@ -1029,7 +1046,11 @@ msgstr ""
msgid "Cantonese"
msgstr ""
#: buzz/model_loader.py:548
#: buzz/transcriber/recording_transcriber.py:338
msgid "Starting Whisper.cpp..."
msgstr ""
#: buzz/model_loader.py:538
msgid "A connection error occurred"
msgstr ""

12
buzz/model_loader.py
View file

@ -7,7 +7,6 @@ import threading
import shutil
import subprocess
import sys
import tempfile
import warnings
import platform
import requests
@ -23,16 +22,6 @@ from huggingface_hub.errors import LocalEntryNotFoundError
from buzz.locale import _
# Catch exception from whisper.dll not getting loaded.
# TODO: Remove flag and try-except when issue with loading
# the DLL in some envs is fixed.
LOADED_WHISPER_CPP_BINARY = False
try:
import buzz.whisper_cpp as whisper_cpp # noqa: F401
LOADED_WHISPER_CPP_BINARY = True
except ImportError:
logging.exception("")
model_root_dir = user_cache_dir("Buzz")
model_root_dir = os.path.join(model_root_dir, "models")
@ -89,6 +78,7 @@ class ModelType(enum.Enum):
)
def is_available(self):
from buzz.transcriber.whisper_cpp import LOADED_WHISPER_CPP_BINARY
if (
# Hide Whisper.cpp option if whisper.dll did not load correctly.
# See: https://github.com/chidiwilliams/buzz/issues/274,

15
buzz/transcriber/file_transcriber.py
View file

@ -6,11 +6,13 @@ import shutil
import tempfile
from abc import abstractmethod
from typing import Optional, List
from pathlib import Path
from PyQt6.QtCore import QObject, pyqtSignal, pyqtSlot
from yt_dlp import YoutubeDL
from buzz.whisper_audio import SAMPLE_RATE
from buzz import whisper_audio
from buzz.assets import APP_BASE_DIR
from buzz.transcriber.transcriber import (
FileTranscriptionTask,
get_output_file_path,
@ -18,6 +20,9 @@ from buzz.transcriber.transcriber import (
OutputFormat,
)
app_env = os.environ.copy()
app_env['PATH'] = os.pathsep.join([os.path.join(APP_BASE_DIR, "_internal")] + [app_env['PATH']])
class FileTranscriber(QObject):
transcription_task: FileTranscriptionTask
@ -35,6 +40,7 @@ class FileTranscriber(QObject):
if self.transcription_task.source == FileTranscriptionTask.Source.URL_IMPORT:
temp_output_path = tempfile.mktemp()
wav_file = temp_output_path + ".wav"
wav_file = str(Path(wav_file).resolve())
cookiefile = os.getenv("BUZZ_DOWNLOAD_COOKIEFILE")
@ -64,16 +70,17 @@ class FileTranscriber(QObject):
"-threads", "0",
"-i", temp_output_path,
"-ac", "1",
"-ar", str(SAMPLE_RATE),
"-ar", str(whisper_audio.SAMPLE_RATE),
"-acodec", "pcm_s16le",
"-loglevel", "panic",
wav_file]
wav_file
]
if sys.platform == "win32":
si = subprocess.STARTUPINFO()
si.dwFlags |= subprocess.STARTF_USESHOWWINDOW
si.wShowWindow = subprocess.SW_HIDE
result = subprocess.run(cmd, capture_output=True, startupinfo=si)
result = subprocess.run(cmd, capture_output=True, startupinfo=si, env=app_env)
else:
result = subprocess.run(cmd, capture_output=True)

87
buzz/transcriber/local_whisper_cpp_server_transcriber.py Normal file
View file

@ -0,0 +1,87 @@
import logging
import os
import time
import subprocess
from typing import Optional, List
from PyQt6.QtCore import QObject
from openai import OpenAI
from buzz.locale import _
from buzz.assets import APP_BASE_DIR
from buzz.transcriber.openai_whisper_api_file_transcriber import OpenAIWhisperAPIFileTranscriber
from buzz.transcriber.transcriber import FileTranscriptionTask, Segment
class LocalWhisperCppServerTranscriber(OpenAIWhisperAPIFileTranscriber):
# To be used on Windows only
def __init__(self, task: FileTranscriptionTask, parent: Optional["QObject"] = None) -> None:
super().__init__(task=task, parent=parent)
self.process = None
self.initialization_error = None
command = [
os.path.join(APP_BASE_DIR, "whisper-server.exe"),
"--port", "3000",
"--inference-path", "/audio/transcriptions",
"--threads", str(os.getenv("BUZZ_WHISPERCPP_N_THREADS", (os.cpu_count() or 8)//2)),
"--language", task.transcription_options.language,
"--model", task.model_path
]
logging.debug(f"Starting Whisper server with command: {' '.join(command)}")
self.process = subprocess.Popen(
command,
stdout=subprocess.DEVNULL, # For debug set to subprocess.PIPE, but it will freeze on Windows after ~30 seconds
stderr=subprocess.PIPE,
shell=False,
creationflags=subprocess.CREATE_NO_WINDOW
)
# Wait for server to start and load model
time.sleep(10)
if self.process is not None and self.process.poll() is None:
logging.debug(f"Whisper server started successfully.")
logging.debug(f"Model: {task.model_path}")
else:
stderr_output = self.process.stderr.read().decode()
logging.error(f"Whisper server failed to start. Error: {stderr_output}")
self.initialization_error = _("Whisper server failed to start. Check logs for details.")
if "ErrorOutOfDeviceMemory" in stderr_output:
self.initialization_error = _("Whisper server failed to start due to insufficient memory. "
"Please try again with a smaller model. "
"To force CPU mode use BUZZ_FORCE_CPU=TRUE environment variable.")
return
self.openai_client = OpenAI(
api_key="not-used",
base_url="http://127.0.0.1:3000"
)
def transcribe(self) -> List[Segment]:
if self.initialization_error:
raise Exception(self.initialization_error)
return super().transcribe()
def stop(self):
if self.process and self.process.poll() is None:
try:
self.process.terminate()
self.process.wait(timeout=5)
except subprocess.TimeoutExpired:
# Force kill if terminate doesn't work within 5 seconds
logging.warning("Whisper server didn't terminate gracefully, force killing")
self.process.kill()
try:
self.process.wait(timeout=2)
except subprocess.TimeoutExpired:
logging.error("Failed to kill whisper server process")
except Exception as e:
logging.error(f"Error stopping whisper server: {e}")
def __del__(self):
self.stop()

14
buzz/transcriber/openai_whisper_api_file_transcriber.py
View file

@ -4,6 +4,8 @@ import os
import sys
import subprocess
import tempfile
from pathlib import Path
from typing import Optional, List
from PyQt6.QtCore import QObject
@ -11,7 +13,7 @@ from openai import OpenAI
from buzz.settings.settings import Settings
from buzz.model_loader import get_custom_api_whisper_model
from buzz.transcriber.file_transcriber import FileTranscriber
from buzz.transcriber.file_transcriber import FileTranscriber, app_env
from buzz.transcriber.transcriber import FileTranscriptionTask, Segment, Task
from buzz.transcriber.whisper_cpp import append_segment
@ -41,6 +43,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
)
mp3_file = tempfile.mktemp() + ".mp3"
mp3_file = str(Path(mp3_file).resolve())
cmd = [
"ffmpeg",
@ -53,7 +56,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
si = subprocess.STARTUPINFO()
si.dwFlags |= subprocess.STARTF_USESHOWWINDOW
si.wShowWindow = subprocess.SW_HIDE
result = subprocess.run(cmd, capture_output=True, startupinfo=si)
result = subprocess.run(cmd, capture_output=True, startupinfo=si, env=app_env)
else:
result = subprocess.run(cmd, capture_output=True)
@ -80,7 +83,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
si.wShowWindow = subprocess.SW_HIDE
duration_secs = float(
subprocess.run(cmd, capture_output=True, check=True, startupinfo=si).stdout.decode("utf-8")
subprocess.run(cmd, capture_output=True, check=True, startupinfo=si, env=app_env).stdout.decode("utf-8")
)
else:
duration_secs = float(
@ -107,6 +110,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
chunk_end = min((i + 1) * chunk_duration, duration_secs)
chunk_file = tempfile.mktemp() + ".mp3"
chunk_file = str(Path(chunk_file).resolve())
# fmt: off
cmd = [
@ -122,7 +126,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
si = subprocess.STARTUPINFO()
si.dwFlags |= subprocess.STARTF_USESHOWWINDOW
si.wShowWindow = subprocess.SW_HIDE
subprocess.run(cmd, capture_output=True, check=True, startupinfo=si)
subprocess.run(cmd, capture_output=True, check=True, startupinfo=si, env=app_env)
else:
subprocess.run(cmd, capture_output=True, check=True)
@ -168,7 +172,7 @@ class OpenAIWhisperAPIFileTranscriber(FileTranscriber):
segments = getattr(transcript, "segments", None)
words = getattr(transcript, "words", None)
if "words" is None and "words" in transcript.model_extra:
if words is None and "words" in transcript.model_extra:
words = transcript.model_extra["words"]
if segments is None:

113
buzz/transcriber/recording_transcriber.py
View file

@ -2,10 +2,12 @@ import datetime
import logging
import platform
import os
import sys
import wave
import time
import tempfile
import threading
import subprocess
from typing import Optional
from platformdirs import user_cache_dir
@ -17,7 +19,9 @@ from openai import OpenAI
from PyQt6.QtCore import QObject, pyqtSignal
from buzz import whisper_audio
from buzz.model_loader import WhisperModelSize, ModelType, get_custom_api_whisper_model
from buzz.locale import _
from buzz.assets import APP_BASE_DIR
from buzz.model_loader import ModelType, get_custom_api_whisper_model
from buzz.settings.settings import Settings
from buzz.transcriber.transcriber import TranscriptionOptions, Task
from buzz.transcriber.whisper_cpp import WhisperCpp
@ -65,8 +69,11 @@ class RecordingTranscriber(QObject):
self.sounddevice = sounddevice
self.openai_client = None
self.whisper_api_model = get_custom_api_whisper_model("")
self.is_windows = sys.platform == "win32"
self.process = None
def start(self):
model = None
model_path = self.model_path
keep_samples = int(self.keep_sample_seconds * self.sample_rate)
@ -80,7 +87,11 @@ class RecordingTranscriber(QObject):
device = "cuda" if use_cuda else "cpu"
model = whisper.load_model(model_path, device=device)
elif self.transcription_options.model.model_type == ModelType.WHISPER_CPP:
model = WhisperCpp(model_path)
# As DLL mode on Windows is somewhat unreliable, will use local whisper-server
if self.is_windows:
self.start_local_whisper_server()
else:
model = WhisperCpp(model_path)
elif self.transcription_options.model.model_type == ModelType.FASTER_WHISPER:
model_root_dir = user_cache_dir("Buzz")
model_root_dir = os.path.join(model_root_dir, "models")
@ -91,6 +102,10 @@ class RecordingTranscriber(QObject):
logging.debug("Unsupported CUDA version (<12), using CPU")
device = "cpu"
if not torch.cuda.is_available():
logging.debug("CUDA is not available, using CPU")
device = "cpu"
if force_cpu != "false":
device = "cpu"
@ -134,7 +149,6 @@ class RecordingTranscriber(QObject):
)
self.is_running = True
amplitude = 0.0
try:
with self.sounddevice.InputStream(
samplerate=self.sample_rate,
@ -159,7 +173,7 @@ class RecordingTranscriber(QObject):
amplitude,
)
if amplitude < 0.01:
if amplitude < 0.025:
time.sleep(0.5)
continue
@ -181,6 +195,8 @@ class RecordingTranscriber(QObject):
elif (
self.transcription_options.model.model_type
== ModelType.WHISPER_CPP
# On Windows we use the local whisper server via OpenAI API
and not self.is_windows
):
assert isinstance(model, WhisperCpp)
result = model.transcribe(
@ -220,7 +236,11 @@ class RecordingTranscriber(QObject):
task=self.transcription_options.task.value,
)
else: # OPEN_AI_WHISPER_API
assert self.openai_client is not None
if self.openai_client is None:
self.transcription.emit(_("A connection error occurred"))
self.stop_recording()
return
# scale samples to 16-bit PCM
pcm_data = (samples * 32767).astype(np.int16).tobytes()
@ -237,7 +257,7 @@ class RecordingTranscriber(QObject):
options = {
"model": self.whisper_api_model,
"file": temp_file,
"response_format": "verbose_json",
"response_format": "json",
"prompt": self.transcription_options.initial_prompt,
}
@ -251,10 +271,17 @@ class RecordingTranscriber(QObject):
else self.openai_client.audio.translations.create(**options)
)
result = {"text": " ".join(
[segment["text"] for segment in transcript.model_extra["segments"]])}
if "segments" in transcript.model_extra:
result = {"text": " ".join(
[segment["text"] for segment in transcript.model_extra["segments"]])}
else:
result = {"text": transcript.text}
except Exception as e:
result = {"text": f"Error: {str(e)}"}
if self.is_running:
result = {"text": f"Error: {str(e)}"}
else:
result = {"text": ""}
os.unlink(temp_filename)
@ -279,6 +306,12 @@ class RecordingTranscriber(QObject):
self.finished.emit()
# Cleanup
if model:
del model
if torch.cuda.is_available():
torch.cuda.empty_cache()
@staticmethod
def get_device_sample_rate(device_id: Optional[int]) -> int:
"""Returns the sample rate to be used for recording. It uses the default sample rate
@ -308,3 +341,65 @@ class RecordingTranscriber(QObject):
def stop_recording(self):
self.is_running = False
if self.process and self.process.poll() is None:
self.process.terminate()
self.process.wait()
def start_local_whisper_server(self):
self.transcription.emit(_("Starting Whisper.cpp..."))
self.process = None
command = [
os.path.join(APP_BASE_DIR, "whisper-server.exe"),
"--port", "3004",
"--inference-path", "/audio/transcriptions",
"--threads", str(os.getenv("BUZZ_WHISPERCPP_N_THREADS", (os.cpu_count() or 8)//2)),
"--language", self.transcription_options.language,
"--model", self.model_path,
"--no-timestamps",
"--no-context", # on Windows context causes duplications of last message
]
logging.debug(f"Starting Whisper server with command: {' '.join(command)}")
self.process = subprocess.Popen(
command,
stdout=subprocess.DEVNULL, # For debug set to subprocess.PIPE, but it will freeze on Windows after ~30 seconds
stderr=subprocess.PIPE,
shell=False,
creationflags=subprocess.CREATE_NO_WINDOW
)
# Wait for server to start and load model
time.sleep(10)
if self.process is not None and self.process.poll() is None:
logging.debug(f"Whisper server started successfully.")
logging.debug(f"Model: {self.model_path}")
else:
stderr_output = self.process.stderr.read().decode()
logging.error(f"Whisper server failed to start. Error: {stderr_output}")
self.transcription.emit(_("Whisper server failed to start. Check logs for details."))
if "ErrorOutOfDeviceMemory" in stderr_output:
message = _("Whisper server failed to start due to insufficient memory. "
"Please try again with a smaller model. "
"To force CPU mode use BUZZ_FORCE_CPU=TRUE environment variable.")
logging.error(message)
self.transcription.emit(message)
self.transcription.emit(_("Whisper server failed to start. Check logs for details."))
return
self.openai_client = OpenAI(
api_key="not-used",
base_url="http://127.0.0.1:3004",
timeout=10.0,
max_retries=0
)
def __del__(self):
if self.process and self.process.poll() is None:
self.process.terminate()
self.process.wait()

2
buzz/transcriber/transcriber.py
View file

@ -21,7 +21,7 @@ class Task(enum.Enum):
TASK_LABEL_TRANSLATIONS = {
Task.TRANSLATE: _("Translate"),
Task.TRANSLATE: _("Translate to English"),
Task.TRANSCRIBE: _("Transcribe"),
}

83
buzz/transcriber/whisper_cpp.py
View file

@ -1,5 +1,6 @@
import platform
import os
import sys
import ctypes
import logging
from typing import Union, Any, List
@ -7,21 +8,43 @@ from typing import Union, Any, List
import numpy as np
from buzz import whisper_audio
from buzz.model_loader import LOADED_WHISPER_CPP_BINARY
from buzz.transcriber.transcriber import Segment, Task, TranscriptionOptions
if LOADED_WHISPER_CPP_BINARY:
from buzz import whisper_cpp
LOADED_WHISPER_CPP_BINARY = False
IS_VULKAN_SUPPORTED = False
try:
import vulkan
IS_COREML_SUPPORTED = False
if platform.system() == "Darwin" and platform.machine() == "arm64":
instance = vulkan.vkCreateInstance(vulkan.VkInstanceCreateInfo(), None)
vulkan.vkDestroyInstance(instance, None)
vulkan_version = vulkan.vkEnumerateInstanceVersion()
major = (vulkan_version >> 22) & 0x3FF
minor = (vulkan_version >> 12) & 0x3FF
logging.debug("Vulkan version = %s.%s", major, minor)
# On macOS, default whisper_cpp is compiled with CoreML (Apple Silicon) or Vulkan (Intel).
# On Windows whisper-server.exe subprocess will be used
if (platform.system() == "Linux") and ((major > 1) or (major == 1 and minor >= 2)):
from buzz.whisper_cpp_vulkan import whisper_cpp_vulkan
IS_VULKAN_SUPPORTED = True
LOADED_WHISPER_CPP_BINARY = True
except (ImportError, Exception) as e:
logging.debug(f"Vulkan import error: {e}")
IS_VULKAN_SUPPORTED = False
if not IS_VULKAN_SUPPORTED:
try:
from buzz import whisper_cpp_coreml # noqa: F401
from buzz.whisper_cpp import whisper_cpp # noqa: F401
IS_COREML_SUPPORTED = True
except ImportError:
logging.exception("")
LOADED_WHISPER_CPP_BINARY = True
except ImportError as e:
logging.exception("whisper_cpp load error: %s", e)
def append_segment(result, txt: bytes, start: int, end: int):
if txt == b'':
@ -45,14 +68,9 @@ def append_segment(result, txt: bytes, start: int, end: int):
class WhisperCpp:
def __init__(self, model: str) -> None:
self.is_coreml_supported = IS_COREML_SUPPORTED
self.is_vulkan_supported = IS_VULKAN_SUPPORTED
if self.is_coreml_supported:
coreml_model = model.replace(".bin", "-encoder.mlmodelc")
if not os.path.exists(coreml_model):
self.is_coreml_supported = False
logging.debug(f"WhisperCpp model {model}, (Core ML: {self.is_coreml_supported})")
logging.debug(f"WhisperCpp model {model}, Vulkan: {self.is_vulkan_supported})")
self.instance = self.get_instance()
self.ctx = self.instance.init_from_file(model)
@ -120,8 +138,8 @@ class WhisperCpp:
}
def get_instance(self):
if self.is_coreml_supported:
return WhisperCppCoreML()
if self.is_vulkan_supported:
return WhisperCppVulkan()
return WhisperCppCpu()
def get_params(
@ -130,7 +148,7 @@ class WhisperCpp:
print_realtime=False,
print_progress=False,
):
params = self.instance.full_default_params(whisper_cpp.WHISPER_SAMPLING_GREEDY)
params = self.instance.full_default_params(0) # WHISPER_SAMPLING_GREEDY
params.n_threads = int(os.getenv("BUZZ_WHISPERCPP_N_THREADS", (os.cpu_count() or 8)//2))
params.print_realtime = print_realtime
params.print_progress = print_progress
@ -216,38 +234,41 @@ class WhisperCppCpu(WhisperCppInterface):
def free(self, ctx):
if ctx and whisper_cpp is not None:
return whisper_cpp.whisper_free(ctx)
return None
class WhisperCppCoreML(WhisperCppInterface):
class WhisperCppVulkan(WhisperCppInterface):
def full_default_params(self, sampling: int):
return whisper_cpp_coreml.whisper_full_default_params(sampling)
return whisper_cpp_vulkan.whisper_full_default_params(sampling)
def get_string(self, string: str):
return whisper_cpp_coreml.String(string.encode())
return whisper_cpp_vulkan.String(string.encode())
def get_encoder_begin_callback(self, callback):
return whisper_cpp_coreml.whisper_encoder_begin_callback(callback)
return whisper_cpp_vulkan.whisper_encoder_begin_callback(callback)
def get_new_segment_callback(self, callback):
return whisper_cpp_coreml.whisper_new_segment_callback(callback)
return whisper_cpp_vulkan.whisper_new_segment_callback(callback)
def init_from_file(self, model: str):
return whisper_cpp_coreml.whisper_init_from_file(model.encode())
return whisper_cpp_vulkan.whisper_init_from_file(model.encode())
def full(self, ctx, params, audio, length):
return whisper_cpp_coreml.whisper_full(ctx, params, audio, length)
return whisper_cpp_vulkan.whisper_full(ctx, params, audio, length)
def full_n_segments(self, ctx):
return whisper_cpp_coreml.whisper_full_n_segments(ctx)
return whisper_cpp_vulkan.whisper_full_n_segments(ctx)
def full_get_segment_text(self, ctx, i):
return whisper_cpp_coreml.whisper_full_get_segment_text(ctx, i)
return whisper_cpp_vulkan.whisper_full_get_segment_text(ctx, i)
def full_get_segment_t0(self, ctx, i):
return whisper_cpp_coreml.whisper_full_get_segment_t0(ctx, i)
return whisper_cpp_vulkan.whisper_full_get_segment_t0(ctx, i)
def full_get_segment_t1(self, ctx, i):
return whisper_cpp_coreml.whisper_full_get_segment_t1(ctx, i)
return whisper_cpp_vulkan.whisper_full_get_segment_t1(ctx, i)
def free(self, ctx):
return whisper_cpp_coreml.whisper_free(ctx)
if ctx and whisper_cpp_vulkan is not None:
return whisper_cpp_vulkan.whisper_free(ctx)
return None

69
buzz/transcriber/whisper_cpp_file_transcriber.py
View file

@ -1,6 +1,7 @@
import ctypes
import logging
import sys
import os
from typing import Optional, List
from PyQt6.QtCore import QObject
@ -9,6 +10,7 @@ from buzz import whisper_audio
from buzz.transcriber.file_transcriber import FileTranscriber
from buzz.transcriber.transcriber import FileTranscriptionTask, Segment, Stopped
from buzz.transcriber.whisper_cpp import WhisperCpp
from buzz.transcriber.local_whisper_cpp_server_transcriber import LocalWhisperCppServerTranscriber
class WhisperCppFileTranscriber(FileTranscriber):
@ -25,7 +27,16 @@ class WhisperCppFileTranscriber(FileTranscriber):
self.transcription_options = task.transcription_options
self.model_path = task.model_path
self.model = WhisperCpp(model=self.model_path)
self.transcriber = None
self.model = None
is_windows = sys.platform == "win32"
force_cpu = os.getenv("BUZZ_FORCE_CPU", "false")
# As DLL mode on Windows is somewhat unreliable, will use local whisper-server
if is_windows and force_cpu == "false":
self.transcriber = LocalWhisperCppServerTranscriber(task, parent)
else:
self.model = WhisperCpp(model=self.model_path)
self.state = self.State()
def transcribe(self) -> List[Segment]:
@ -41,32 +52,42 @@ class WhisperCppFileTranscriber(FileTranscriber):
self.transcription_options.word_level_timings,
)
audio = whisper_audio.load_audio(self.transcription_task.file_path)
self.duration_audio_ms = len(audio) * 1000 / whisper_audio.SAMPLE_RATE
if self.model:
audio = whisper_audio.load_audio(self.transcription_task.file_path)
self.duration_audio_ms = len(audio) * 1000 / whisper_audio.SAMPLE_RATE
whisper_params = self.model.get_params(
transcription_options=self.transcription_options
)
whisper_params.encoder_begin_callback_user_data = ctypes.c_void_p(
id(self.state)
)
whisper_params.encoder_begin_callback = (
self.model.get_instance().get_encoder_begin_callback(self.encoder_begin_callback)
)
whisper_params.new_segment_callback_user_data = ctypes.c_void_p(id(self.state))
whisper_params.new_segment_callback = self.model.get_instance().get_new_segment_callback(
self.new_segment_callback
)
whisper_params = self.model.get_params(
transcription_options=self.transcription_options
)
whisper_params.encoder_begin_callback_user_data = ctypes.c_void_p(
id(self.state)
)
whisper_params.encoder_begin_callback = (
self.model.get_instance().get_encoder_begin_callback(self.encoder_begin_callback)
)
whisper_params.new_segment_callback_user_data = ctypes.c_void_p(id(self.state))
whisper_params.new_segment_callback = self.model.get_instance().get_new_segment_callback(
self.new_segment_callback
)
result = self.model.transcribe(
audio=self.transcription_task.file_path, params=whisper_params
)
result = self.model.transcribe(
audio=self.transcription_task.file_path, params=whisper_params
)
if not self.state.running:
raise Stopped
self.state.running = False
return result["segments"]
# On Windows we use the local whisper server
if self.transcriber is not None:
return self.transcriber.transcribe()
if not self.state.running:
raise Stopped
self.state.running = False
return result["segments"]
return []
def new_segment_callback(self, ctx, _state, _n_new, user_data):
n_segments = self.model.get_instance().full_n_segments(ctx)
@ -89,3 +110,9 @@ class WhisperCppFileTranscriber(FileTranscriber):
def stop(self):
self.state.running = False
if self.transcriber is not None:
self.transcriber.stop()
def __del__(self):
self.stop()

52
buzz/transcriber/whisper_file_transcriber.py
View file

@ -46,6 +46,8 @@ class WhisperFileTranscriber(FileTranscriber):
self.segments = []
self.started_process = False
self.stopped = False
self.recv_pipe = None
self.send_pipe = None
def transcribe(self) -> List[Segment]:
time_started = datetime.datetime.now()
@ -56,24 +58,30 @@ class WhisperFileTranscriber(FileTranscriber):
if torch.cuda.is_available():
logging.debug(f"CUDA version detected: {torch.version.cuda}")
recv_pipe, send_pipe = multiprocessing.Pipe(duplex=False)
self.recv_pipe, self.send_pipe = multiprocessing.Pipe(duplex=False)
self.current_process = multiprocessing.Process(
target=self.transcribe_whisper, args=(send_pipe, self.transcription_task)
target=self.transcribe_whisper, args=(self.send_pipe, self.transcription_task)
)
if not self.stopped:
self.current_process.start()
self.started_process = True
self.read_line_thread = Thread(target=self.read_line, args=(recv_pipe,))
self.read_line_thread = Thread(target=self.read_line, args=(self.recv_pipe,))
self.read_line_thread.start()
self.current_process.join()
if self.current_process.exitcode != 0:
send_pipe.close()
self.send_pipe.close()
self.read_line_thread.join()
# Join read_line_thread with timeout to prevent hanging
if self.read_line_thread and self.read_line_thread.is_alive():
self.read_line_thread.join(timeout=3)
if self.read_line_thread.is_alive():
logging.warning("Read line thread didn't terminate gracefully in transcribe()")
self.started_process = False
logging.debug(
"whisper process completed with code = %s, time taken = %s,"
@ -153,6 +161,10 @@ class WhisperFileTranscriber(FileTranscriber):
logging.debug("Unsupported CUDA version (<12), using CPU")
device = "cpu"
if not torch.cuda.is_available():
logging.debug("CUDA is not available, using CPU")
device = "cpu"
if force_cpu != "false":
device = "cpu"
@ -168,7 +180,8 @@ class WhisperFileTranscriber(FileTranscriber):
audio=task.file_path,
language=task.transcription_options.language,
task=task.transcription_options.task.value,
temperature=task.transcription_options.temperature,
# Prevent crash on Windows https://github.com/SYSTRAN/faster-whisper/issues/71#issuecomment-1526263764
temperature = 0 if platform.system() == "Windows" else task.transcription_options.temperature,
initial_prompt=task.transcription_options.initial_prompt,
word_timestamps=task.transcription_options.word_level_timings,
no_speech_threshold=0.4,
@ -249,8 +262,30 @@ class WhisperFileTranscriber(FileTranscriber):
def stop(self):
self.stopped = True
if self.started_process:
self.current_process.terminate()
# Use timeout to avoid hanging indefinitely
self.current_process.join(timeout=5)
if self.current_process.is_alive():
logging.warning("Process didn't terminate gracefully, force killing")
self.current_process.kill()
self.current_process.join(timeout=2)
# Close pipes to unblock the read_line thread
try:
if hasattr(self, 'send_pipe'):
self.send_pipe.close()
if hasattr(self, 'recv_pipe'):
self.recv_pipe.close()
except Exception as e:
logging.debug(f"Error closing pipes: {e}")
# Join read_line_thread with timeout to prevent hanging
if self.read_line_thread and self.read_line_thread.is_alive():
self.read_line_thread.join(timeout=3)
if self.read_line_thread.is_alive():
logging.warning("Read line thread didn't terminate gracefully")
def read_line(self, pipe: Connection):
while True:
@ -260,7 +295,10 @@ class WhisperFileTranscriber(FileTranscriber):
# Uncomment to debug
# print(f"*** DEBUG ***: {line}")
except EOFError: # Connection closed
except (EOFError, BrokenPipeError, ConnectionResetError): # Connection closed or broken
break
except Exception as e:
logging.debug(f"Error reading from pipe: {e}")
break
if line == self.READ_LINE_THREAD_STOP_TOKEN:

38
buzz/transformers_whisper.py
View file

@ -17,7 +17,8 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
inputs_len = inputs.shape[0]
step = chunk_len - stride_left - stride_right
for chunk_start_idx in range(0, inputs_len, step):
# Print progress to stderr
# Buzz will print progress to stderr
progress = int((chunk_start_idx / inputs_len) * 100)
sys.stderr.write(f"{progress}%\n")
@ -27,8 +28,7 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
if dtype is not None:
processed = processed.to(dtype=dtype)
_stride_left = 0 if chunk_start_idx == 0 else stride_left
# all right strides must be full, otherwise it is the last item
is_last = chunk_end_idx > inputs_len if stride_right > 0 else chunk_end_idx >= inputs_len
is_last = chunk_end_idx >= inputs_len
_stride_right = 0 if is_last else stride_right
chunk_len = chunk.shape[0]
@ -98,7 +98,7 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
# of the original length in the stride so we can cut properly.
stride = (inputs.shape[0], int(round(stride[0] * ratio)), int(round(stride[1] * ratio)))
if not isinstance(inputs, np.ndarray):
raise ValueError(f"We expect a numpy ndarray as input, got `{type(inputs)}`")
raise TypeError(f"We expect a numpy ndarray as input, got `{type(inputs)}`")
if len(inputs.shape) != 1:
raise ValueError("We expect a single channel audio input for AutomaticSpeechRecognitionPipeline")
@ -109,7 +109,7 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
if isinstance(stride_length_s, (int, float)):
stride_length_s = [stride_length_s, stride_length_s]
# XXX: Carefuly, this variable will not exist in `seq2seq` setting.
# XXX: Carefully, this variable will not exist in `seq2seq` setting.
# Currently chunking is not possible at this level for `seq2seq` so
# it's ok.
align_to = getattr(self.model.config, "inputs_to_logits_ratio", 1)
@ -120,11 +120,11 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
if chunk_len < stride_left + stride_right:
raise ValueError("Chunk length must be superior to stride length")
# Will use our custom chunk_iter with progress
# Buzz use our custom chunk_iter with progress
for item in self.chunk_iter(
inputs, self.feature_extractor, chunk_len, stride_left, stride_right, self.torch_dtype
):
yield item
yield {**item, **extra}
else:
if self.type == "seq2seq_whisper" and inputs.shape[0] > self.feature_extractor.n_samples:
processed = self.feature_extractor(
@ -133,12 +133,25 @@ class PipelineWithProgress(AutomaticSpeechRecognitionPipeline): # pragma: no co
truncation=False,
padding="longest",
return_tensors="pt",
return_attention_mask=True,
)
else:
processed = self.feature_extractor(
inputs, sampling_rate=self.feature_extractor.sampling_rate, return_tensors="pt"
)
if self.type == "seq2seq_whisper" and stride is None:
processed = self.feature_extractor(
inputs,
sampling_rate=self.feature_extractor.sampling_rate,
return_tensors="pt",
return_token_timestamps=True,
return_attention_mask=True,
)
extra["num_frames"] = processed.pop("num_frames")
else:
processed = self.feature_extractor(
inputs,
sampling_rate=self.feature_extractor.sampling_rate,
return_tensors="pt",
return_attention_mask=True,
)
if self.torch_dtype is not None:
processed = processed.to(dtype=self.torch_dtype)
if stride is not None:
@ -193,7 +206,8 @@ class TransformersWhisper:
chunk_length_s=30 if word_timestamps else None,
torch_dtype=torch_dtype,
device=device,
)
ignore_warning=True # Ignore warning about chunk_length_s being experimental for seq2seq models
)
transcript = pipe(
audio,

2
buzz/widgets/main_window.py
View file

@ -430,6 +430,8 @@ class MainWindow(QMainWindow):
if self.transcription_viewer_widget is not None:
self.transcription_viewer_widget.close()
logging.debug("Closing MainWindow")
super().closeEvent(event)
def save_geometry(self):

4
buzz/widgets/preferences_dialog/general_preferences_widget.py
View file

@ -187,7 +187,7 @@ class GeneralPreferencesWidget(QWidget):
def on_click_test_openai_api_key_button(self):
self.test_openai_api_key_button.setEnabled(False)
job = TestOpenAIApiKeyJob(api_key=self.openai_api_key)
job = ValidateOpenAIApiKeyJob(api_key=self.openai_api_key)
job.signals.success.connect(self.on_test_openai_api_key_success)
job.signals.failed.connect(self.on_test_openai_api_key_failure)
job.setAutoDelete(True)
@ -265,7 +265,7 @@ class GeneralPreferencesWidget(QWidget):
def on_recording_transcriber_mode_changed(self, value):
self.settings.set_value(Settings.Key.RECORDING_TRANSCRIBER_MODE, value)
class TestOpenAIApiKeyJob(QRunnable):
class ValidateOpenAIApiKeyJob(QRunnable):
class Signals(QObject):
success = pyqtSignal()
failed = pyqtSignal(str)

4
buzz/widgets/recording_transcriber_widget.py
View file

@ -306,7 +306,9 @@ class RecordingTranscriberWidget(QWidget):
self.reset_recording_controls()
self.model_loader = None
if model_path == "":
if model_path == "" and self.transcription_options.model.model_type != ModelType.OPEN_AI_WHISPER_API:
self.on_transcriber_error("")
logging.error("Model path is empty, cannot start recording.")
return
self.transcription_thread = QThread()

6
dll_backup/Readme.md
View file

@ -1,6 +0,0 @@
DLLs extracted from whisper.cpp release v1.6.2
https://github.com/ggerganov/whisper.cpp/actions/runs/9251036100
https://github.com/ggerganov/whisper.cpp/commit/c7b6988678779901d02ceba1a8212d2c9908956e
whisper.cpp submodule commit version should match the commit where the
whisper.dll are extracted from to ensure compatibility.

BIN
dll_backup/whisper.dll
View file

Binary file not shown.

BIN
dll_backup/win32/whisper.dll
View file

Binary file not shown.

33
poetry.lock generated
View file

@ -1181,7 +1181,6 @@ description = "Fast transfer of large files with the Hugging Face Hub."
optional = false
python-versions = ">=3.8"
groups = ["main"]
markers = "platform_machine == \"x86_64\" or platform_machine == \"amd64\" or platform_machine == \"arm64\" or platform_machine == \"aarch64\""
files = [
{file = "hf_xet-1.1.5-cp37-abi3-macosx_10_12_x86_64.whl", hash = "sha256:f52c2fa3635b8c37c7764d8796dfa72706cc4eded19d638331161e82b0792e23"},
{file = "hf_xet-1.1.5-cp37-abi3-macosx_11_0_arm64.whl", hash = "sha256:9fa6e3ee5d61912c4a113e0708eaaef987047616465ac7aa30f7121a48fc1af8"},
@ -3427,6 +3426,21 @@ pytest = "*"
dev = ["pre-commit", "tox"]
doc = ["sphinx", "sphinx-rtd-theme"]
[[package]]
name = "pytest-timeout"
version = "2.4.0"
description = "pytest plugin to abort hanging tests"
optional = false
python-versions = ">=3.7"
groups = ["dev"]
files = [
{file = "pytest_timeout-2.4.0-py3-none-any.whl", hash = "sha256:c42667e5cdadb151aeb5b26d114aff6bdf5a907f176a007a30b940d3d865b5c2"},
{file = "pytest_timeout-2.4.0.tar.gz", hash = "sha256:7e68e90b01f9eff71332b25001f85c75495fc4e3a836701876183c4bcfd0540a"},
]
[package.dependencies]
pytest = ">=7.0.0"
[[package]]
name = "pytest-xvfb"
version = "2.0.0"
@ -4790,6 +4804,21 @@ platformdirs = ">=3.9.1,<5"
docs = ["furo (>=2023.7.26)", "proselint (>=0.13)", "sphinx (>=7.1.2,!=7.3)", "sphinx-argparse (>=0.4)", "sphinxcontrib-towncrier (>=0.2.1a0)", "towncrier (>=23.6)"]
test = ["covdefaults (>=2.3)", "coverage (>=7.2.7)", "coverage-enable-subprocess (>=1)", "flaky (>=3.7)", "packaging (>=23.1)", "pytest (>=7.4)", "pytest-env (>=0.8.2)", "pytest-freezer (>=0.4.8) ; platform_python_implementation == \"PyPy\" or platform_python_implementation == \"CPython\" and sys_platform == \"win32\" and python_version >= \"3.13\"", "pytest-mock (>=3.11.1)", "pytest-randomly (>=3.12)", "pytest-timeout (>=2.1)", "setuptools (>=68)", "time-machine (>=2.10) ; platform_python_implementation == \"CPython\""]
[[package]]
name = "vulkan"
version = "1.3.275.1"
description = "Ultimate Python binding for Vulkan API"
optional = false
python-versions = "*"
groups = ["main"]
files = [
{file = "vulkan-1.3.275.1-py3-none-any.whl", hash = "sha256:e1e0ddf57d3a7d19f79ebf1e192b20dbd378172b027cad4f495d961b51409586"},
{file = "vulkan-1.3.275.1.tar.gz", hash = "sha256:5cf7961cb4a4e60d063eb815147c2289244575b75a4512089cc6e8a959b996d2"},
]
[package.dependencies]
cffi = ">=1.10"
[[package]]
name = "wrapt"
version = "1.17.2"
@ -4925,4 +4954,4 @@ type = ["pytest-mypy"]
[metadata]
lock-version = "2.1"
python-versions = ">=3.9,<3.13"
content-hash = "6fd00715df18f91cd02b7524a34897f408f686eab80fd77b2d8eefd4f2ce99ca"
content-hash = "068de55b5f53b353c9ac9880a1d420f4187decbd547f8b2b871f3dc790a977c8"

10
pyproject.toml
View file

@ -20,12 +20,9 @@ packages = [
{ include = "buzz" },
]
include = [
{ path = "buzz/libwhisper.*", format = ["sdist", "wheel"] },
{ path = "buzz/libwhisper-coreml.*", format = ["sdist", "wheel"] },
{ path = "buzz/*.dll", format = ["sdist", "wheel"] },
{ path = "buzz/whisper_cpp.py", format = ["sdist", "wheel"] },
{ path = "buzz/whisper_cpp/*", format = ["sdist", "wheel"] },
{ path = "buzz/whisper_cpp_vulkan/*", format = ["sdist", "wheel"] },
{ path = "buzz/locale/*/LC_MESSAGES/buzz.mo", format = ["sdist", "wheel"] },
{ path = "buzz/dll_backup/*", format = ["sdist", "wheel"] },
]
[[tool.poetry.source]]
@ -80,6 +77,8 @@ urllib3 = "^2.3.0"
demucs = {url = "https://github.com/raivisdejus/demucs/releases/download/4.1.0a3/demucs-4.1.0a3-py3-none-any.whl"}
posthog = "^3.23.0"
onnxruntime = "^1.18.0"
vulkan = "^1.3.275.1"
hf-xet = "^1.1.5"
[tool.poetry.group.dev.dependencies]
autopep8 = "^1.7.0"
@ -91,6 +90,7 @@ pytest-cov = "^4.0.0"
pytest-qt = "^4.1.0"
pytest-xvfb = "^2.0.0"
pytest-mock = "^3.12.0"
pytest-timeout = "^2.4.0"
pylint = "^2.15.5"
pre-commit = "^2.20.0"
pytest-benchmark = "^4.0.0"

5
pytest.ini
View file

@ -2,5 +2,8 @@
log_cli = 1
log_cli_level = DEBUG
qt_api=pyqt6
log_format = %(asctime)s %(levelname)s %(message)s
log_format = %(asctime)s %(levelname)s %(module)s::%(funcName)s %(message)s
log_date_format = %Y-%m-%d %H:%M:%S
addopts = -x
markers =
timeout: set a timeout on a test function.

67
snap/snapcraft.yaml
View file

@ -1,5 +1,5 @@
# Development notes:
# - To build the snap run `snapcraft clean` and `snapcraft --debug --verbosity verbose`
# - To build the snap run `snapcraft clean` and `snapcraft --verbose`
# - To install local snap `snap install ./buzz_*.snap --dangerous`
name: buzz
base: core22
@ -83,7 +83,12 @@ parts:
plugin: python
source: .
build-packages:
- wget
- portaudio19-dev
- qt6-declarative-dev
- qt6-multimedia-dev
- libvulkan-dev
- cmake
stage-packages:
# Audio
- ffmpeg
@ -100,8 +105,6 @@ parts:
- libgstreamer1.0-0
- libgstreamer-plugins-base1.0-0
- libgstreamer-plugins-good1.0-0
- qt6-declarative-dev
- qt6-multimedia-dev
# Display
- libxkbcommon-x11-0
- libxcb-icccm4
@ -114,7 +117,20 @@ parts:
- libxcb-cursor0
# GPU
- libglu1-mesa
- libvulkan1
- mesa-vulkan-drivers
python-packages:
- ctypesgen
- setuptools
- cmake
- polib
override-build: |
# https://vulkan.lunarg.com/doc/view/latest/linux/getting_started_ubuntu.html
wget -qO- https://packages.lunarg.com/lunarg-signing-key-pub.asc | tee /etc/apt/trusted.gpg.d/lunarg.asc
wget -qO /etc/apt/sources.list.d/lunarg-vulkan-jammy.list http://packages.lunarg.com/vulkan/lunarg-vulkan-jammy.list
apt update
apt install -y vulkan-sdk
# Clear cache to save space on CI
apt clean
@ -122,39 +138,48 @@ parts:
pip install poetry
make translation_mo
pip install .
pip uninstall -y ctranslate2
pip uninstall -y torch torchaudio nvidia-cublas-cu12 nvidia-cuda-cupti-cu12 nvidia-cuda-nvrtc-cu12 nvidia-cuda-runtime-cu12 nvidia-cudnn-cu12 nvidia-cufft-cu12 nvidia-cufile-cu12 nvidia-curand-cu12 nvidia-cusolver-cu12 nvidia-cusparse-cu12 nvidia-cusparselt-cu12 nvidia-nccl-cu12 nvidia-nvjitlink-cu12 nvidia-nvtx-cu12
pip cache purge
pip install https://github.com/raivisdejus/CTranslate2-no-execstack/releases/download/v4.2.1/ctranslate2-4.2.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
pip install -U torch==2.7.1+cu128 torchaudio==2.7.1+cu128 --index-url https://download.pytorch.org/whl/cu128
pip cache purge
pip install nvidia-cublas-cu12==12.8.3.14 nvidia-cuda-cupti-cu12==12.8.57 nvidia-cuda-nvrtc-cu12==12.8.61 --extra-index-url https://pypi.ngc.nvidia.com
pip cache purge
pip install nvidia-cuda-runtime-cu12==12.8.57 nvidia-cudnn-cu12==9.7.1.26 nvidia-cufft-cu12==11.3.3.41 --extra-index-url https://pypi.ngc.nvidia.com
pip cache purge
pip install nvidia-curand-cu12==10.3.9.55 nvidia-cusolver-cu12==11.7.2.55 nvidia-cusparse-cu12==12.5.4.2 --extra-index-url https://pypi.ngc.nvidia.com
pip install nvidia-curand-cu12==10.3.9.55 nvidia-cusolver-cu12==11.7.2.55 nvidia-cusparse-cu12==12.5.7.53 --extra-index-url https://pypi.ngc.nvidia.com
pip cache purge
pip install nvidia-cusparselt-cu12==0.6.3 nvidia-nvjitlink-cu12==12.8.61 nvidia-nvtx-cu12==12.8.55 --extra-index-url https://pypi.ngc.nvidia.com
pip cache purge
python3 build.py
mkdir $CRAFT_PART_INSTALL/buzz
cp $CRAFT_PART_BUILD/buzz/whisper_cpp.py $CRAFT_PART_INSTALL/buzz/
cp $CRAFT_PART_BUILD/buzz/libwhisper.so $CRAFT_PART_INSTALL/buzz/
cp -r $CRAFT_PART_BUILD/buzz/whisper_cpp $CRAFT_PART_INSTALL/buzz/
cp -r $CRAFT_PART_BUILD/buzz/whisper_cpp_vulkan $CRAFT_PART_INSTALL/buzz/
# Create desktop file
mkdir -p $CRAFT_PART_INSTALL/usr/share/applications
cp $CRAFT_PART_BUILD/buzz.desktop $CRAFT_PART_INSTALL/usr/share/applications/
after: [ desktop-qt5 ]
desktop-file:
graphics-core22:
after: [ buzz ]
source: https://github.com/canonical/gpu-snap.git
plugin: dump
source: .
organize:
buzz.desktop: usr/share/applications/buzz.desktop
# To prevent error of the same file having different permissions
override-prime: |
craftctl default
${CRAFT_PART_SRC}/bin/graphics-core22-cleanup mesa-core22 nvidia-core22
stage:
- -buzz/whisper_cpp.py
- ./*
- -venv/**
prime:
- ./*
- -venv/**
- bin/graphics-core22-wrapper
apps:
buzz:
command-chain:
- bin/graphics-core22-wrapper
command: bin/desktop-launch python3 -m buzz
desktop: usr/share/applications/buzz.desktop
environment:
@ -202,7 +227,19 @@ plugs:
interface: content
target: $SNAP/share/sounds
default-provider: gtk-common-themes
graphics-core22:
interface: content
target: $SNAP/graphics
default-provider: mesa-core22
layout:
/usr/lib/$SNAPCRAFT_ARCH_TRIPLET/alsa-lib:
bind: $SNAP/usr/lib/$SNAPCRAFT_ARCH_TRIPLET/alsa-lib
/usr/share/libdrm:
bind: $SNAP/graphics/libdrm
/usr/share/drirc.d:
symlink: $SNAP/graphics/drirc.d
/usr/share/X11/XErrorDB:
symlink: $SNAP/graphics/X11/XErrorDB
/usr/share/X11/locale:
symlink: $SNAP/graphics/X11/locale

42
tests/transcriber/file_transcriber_queue_worker_test.py
View file

@ -2,12 +2,13 @@ import pytest
import unittest.mock
from PyQt6.QtCore import QCoreApplication, QThread
from buzz.file_transcriber_queue_worker import FileTranscriberQueueWorker
from buzz.model_loader import ModelType, TranscriptionModel
from buzz.model_loader import ModelType, TranscriptionModel, WhisperModelSize
from buzz.transcriber.transcriber import FileTranscriptionTask, TranscriptionOptions, FileTranscriptionOptions
from buzz.transcriber.whisper_cpp_file_transcriber import WhisperCppFileTranscriber
from tests.audio import test_multibyte_utf8_audio_path
import time
@pytest.fixture(scope="session")
def qapp():
app = QCoreApplication.instance()
@ -16,6 +17,7 @@ def qapp():
yield app
app.quit()
@pytest.fixture
def worker(qapp):
worker = FileTranscriberQueueWorker()
@ -28,29 +30,27 @@ def worker(qapp):
thread.quit()
thread.wait()
@pytest.fixture
def audio_file():
# Use a small, existing audio file for testing
return test_multibyte_utf8_audio_path
def test_transcription_with_whisper_cpp_tiny_no_speech_extraction(worker, audio_file):
def test_transcription_with_whisper_cpp_tiny_no_speech_extraction(worker):
options = TranscriptionOptions(
model=TranscriptionModel(model_type=ModelType.WHISPER_CPP, whisper_model_size="tiny"),
model=TranscriptionModel(model_type=ModelType.WHISPER_CPP, whisper_model_size=WhisperModelSize.TINY),
extract_speech=False
)
task = FileTranscriptionTask(file_path=str(audio_file), transcription_options=options, file_transcription_options=FileTranscriptionOptions(), model_path="mock_path")
task = FileTranscriptionTask(file_path=str(test_multibyte_utf8_audio_path), transcription_options=options,
file_transcription_options=FileTranscriptionOptions(), model_path="mock_path")
with unittest.mock.patch.object(WhisperCppFileTranscriber, 'run') as mock_run:
with unittest.mock.patch('buzz.transcriber.whisper_cpp_file_transcriber.LocalWhisperCppServerTranscriber'), \
unittest.mock.patch.object(WhisperCppFileTranscriber, 'run') as mock_run:
mock_run.side_effect = lambda: worker.current_transcriber.completed.emit([
{"start": 0, "end": 1000, "text": "Test transcription."}
])
completed_spy = unittest.mock.Mock()
worker.task_completed.connect(completed_spy)
worker.add_task(task)
# Wait for the signal to be emitted
timeout = 5 # seconds
timeout = 10 # seconds
start_time = time.time()
while not completed_spy.called and (time.time() - start_time) < timeout:
QCoreApplication.processEvents()
@ -62,17 +62,19 @@ def test_transcription_with_whisper_cpp_tiny_no_speech_extraction(worker, audio_
assert len(args[1]) > 0
assert args[1][0]["text"] == "Test transcription."
def test_transcription_with_whisper_cpp_tiny_with_speech_extraction(worker, audio_file):
def test_transcription_with_whisper_cpp_tiny_with_speech_extraction(worker):
options = TranscriptionOptions(
model=TranscriptionModel(model_type=ModelType.WHISPER_CPP, whisper_model_size="tiny"),
model=TranscriptionModel(model_type=ModelType.WHISPER_CPP, whisper_model_size=WhisperModelSize.TINY),
extract_speech=True
)
task = FileTranscriptionTask(file_path=str(audio_file), transcription_options=options, file_transcription_options=FileTranscriptionOptions(), model_path="mock_path")
with unittest.mock.patch('demucs.api.Separator') as mock_separator_class, \
unittest.mock.patch('demucs.api.save_audio') as mock_save_audio, \
unittest.mock.patch.object(WhisperCppFileTranscriber, 'run') as mock_run:
task = FileTranscriptionTask(file_path=str(test_multibyte_utf8_audio_path), transcription_options=options,
file_transcription_options=FileTranscriptionOptions(), model_path="mock_path")
with unittest.mock.patch('buzz.transcriber.whisper_cpp_file_transcriber.LocalWhisperCppServerTranscriber'), \
unittest.mock.patch('demucs.api.Separator') as mock_separator_class, \
unittest.mock.patch('demucs.api.save_audio') as mock_save_audio, \
unittest.mock.patch.object(WhisperCppFileTranscriber, 'run') as mock_run:
# Mock demucs.api.Separator and save_audio
mock_separator_instance = unittest.mock.Mock()
mock_separator_instance.separate_audio_file.return_value = (None, {"vocals": "mock_vocals_data"})
@ -88,7 +90,7 @@ def test_transcription_with_whisper_cpp_tiny_with_speech_extraction(worker, audi
worker.add_task(task)
# Wait for the signal to be emitted
timeout = 5 # seconds
timeout = 10 # seconds
start_time = time.time()
while not completed_spy.called and (time.time() - start_time) < timeout:
QCoreApplication.processEvents()
@ -100,4 +102,4 @@ def test_transcription_with_whisper_cpp_tiny_with_speech_extraction(worker, audi
args, kwargs = completed_spy.call_args
assert args[0] == task
assert len(args[1]) > 0
assert args[1][0]["text"] == "Test transcription with speech extraction."
assert args[1][0]["text"] == "Test transcription with speech extraction."

30
tests/transcriber/recording_transcriber_test.py
View file

@ -1,7 +1,12 @@
import os
import sys
import time
from unittest.mock import Mock, patch
from PyQt6.QtCore import QThread
from buzz.locale import _
from buzz.assets import APP_BASE_DIR
from buzz.model_loader import TranscriptionModel, ModelType, WhisperModelSize
from buzz.transcriber.recording_transcriber import RecordingTranscriber
from buzz.transcriber.transcriber import TranscriptionOptions, Task
@ -21,6 +26,10 @@ class TestRecordingTranscriber:
model_path = get_model_path(transcription_model)
model_exe_path = os.path.join(APP_BASE_DIR, "whisper-server.exe")
if sys.platform.startswith("win"):
assert os.path.exists(model_exe_path), f"{model_exe_path} does not exist"
transcriber = RecordingTranscriber(
transcription_options=TranscriptionOptions(
model=transcription_model, language="fr", task=Task.TRANSCRIBE
@ -34,17 +43,24 @@ class TestRecordingTranscriber:
thread.started.connect(transcriber.start)
mock_transcription = Mock()
transcriber.transcription.connect(mock_transcription)
transcriptions = []
with qtbot.wait_signal(transcriber.transcription, timeout=60 * 1000):
thread.start()
def on_transcription(text):
transcriptions.append(text)
transcriber.stop_recording()
transcriber.transcription.connect(on_transcription)
text = mock_transcription.call_args[0][0]
assert "Bienvenue dans Passe" in text
thread.start()
qtbot.waitUntil(lambda: len(transcriptions) == 3, timeout=60_000)
# any string in any transcription
strings_to_check = [_("Starting Whisper.cpp..."), "Bienvenue dans Passe"]
assert any(s in t for s in strings_to_check for t in transcriptions)
# Wait for the thread to finish
transcriber.stop_recording()
time.sleep(10)
thread.quit()
thread.wait()
time.sleep(3)

19
tests/transcriber/whisper_cpp_file_transcriber_test.py
View file

@ -1,5 +1,8 @@
import os
import time
from typing import List
from unittest.mock import Mock
from pathlib import Path
import pytest
from pytestqt.qtbot import QtBot
@ -31,8 +34,9 @@ class TestWhisperCppFileTranscriber:
def test_transcribe(
self, qtbot: QtBot, word_level_timings: bool, expected_segments: List[Segment]
):
os.environ["BUZZ_FORCE_CPU"] = "true"
file_transcription_options = FileTranscriptionOptions(
file_paths=[test_audio_path]
file_paths=[str(Path(test_audio_path).resolve())]
)
transcription_options = TranscriptionOptions(
language="fr",
@ -47,7 +51,7 @@ class TestWhisperCppFileTranscriber:
model_path = get_model_path(transcription_options.model)
transcriber = WhisperCppFileTranscriber(
task=FileTranscriptionTask(
file_path=test_audio_path,
file_path=str(Path(test_audio_path).resolve()),
transcription_options=transcription_options,
file_transcription_options=file_transcription_options,
model_path=model_path,
@ -76,6 +80,9 @@ class TestWhisperCppFileTranscriber:
assert expected_segment.end == segments[i].end
assert expected_segment.text in segments[i].text
transcriber.stop()
time.sleep(3)
@pytest.mark.parametrize(
"word_level_timings,expected_segments",
[
@ -91,8 +98,9 @@ class TestWhisperCppFileTranscriber:
def test_transcribe_latvian(
self, qtbot: QtBot, word_level_timings: bool, expected_segments: List[Segment]
):
os.environ["BUZZ_FORCE_CPU"] = "true"
file_transcription_options = FileTranscriptionOptions(
file_paths=[test_multibyte_utf8_audio_path]
file_paths=[str(Path(test_multibyte_utf8_audio_path).resolve())]
)
transcription_options = TranscriptionOptions(
language="lv",
@ -107,7 +115,7 @@ class TestWhisperCppFileTranscriber:
model_path = get_model_path(transcription_options.model)
transcriber = WhisperCppFileTranscriber(
task=FileTranscriptionTask(
file_path=test_multibyte_utf8_audio_path,
file_path=str(Path(test_multibyte_utf8_audio_path).resolve()),
transcription_options=transcription_options,
file_transcription_options=file_transcription_options,
model_path=model_path,
@ -135,3 +143,6 @@ class TestWhisperCppFileTranscriber:
assert expected_segment.start == segments[i].start
assert expected_segment.end == segments[i].end
assert expected_segment.text in segments[i].text
transcriber.stop()
time.sleep(3)

11
tests/transcriber/whisper_file_transcriber_test.py
View file

@ -230,6 +230,9 @@ class TestWhisperFileTranscriber:
assert len(segments[i].text) > 0
logging.debug(f"{segments[i].start} {segments[i].end} {segments[i].text}")
transcriber.stop()
time.sleep(3)
def test_transcribe_from_url(self, qtbot):
url = (
"https://github.com/chidiwilliams/buzz/raw/main/testdata/whisper-french.mp3"
@ -270,6 +273,9 @@ class TestWhisperFileTranscriber:
assert len(segments[i].text) > 0
logging.debug(f"{segments[i].start} {segments[i].end} {segments[i].text}")
transcriber.stop()
time.sleep(3)
def test_transcribe_from_folder_watch_source(self, qtbot):
file_path = tempfile.mktemp(suffix=".mp3")
shutil.copy(test_audio_path, file_path)
@ -301,6 +307,9 @@ class TestWhisperFileTranscriber:
)
assert len(glob.glob("*.txt", root_dir=output_directory)) > 0
transcriber.stop()
time.sleep(3)
@pytest.mark.skip()
def test_transcribe_stop(self):
output_file_path = os.path.join(tempfile.gettempdir(), "whisper.txt")
@ -335,3 +344,5 @@ class TestWhisperFileTranscriber:
# Assert that file was not created
assert os.path.isfile(output_file_path) is False
time.sleep(3)

45
tests/widgets/main_window_test.py
View file

@ -1,3 +1,4 @@
import logging
import os
from typing import List
from unittest.mock import patch
@ -95,6 +96,7 @@ class TestMainWindow:
window.close()
@pytest.mark.timeout(300)
def test_should_run_and_cancel_transcription_task(
self, qtbot, db, transcription_service
):
@ -105,19 +107,44 @@ class TestMainWindow:
table_widget = self._get_tasks_table(window)
qtbot.wait_until(
self._get_assert_task_status_callback(table_widget, 0, "in_progress"),
timeout=2 * 60 * 1000,
)
try:
qtbot.wait_until(
self._get_assert_task_status_callback(table_widget, 0, "in_progress"),
timeout=60 * 1000,
)
except Exception:
logging.error("Task never reached 'in_progress' status")
assert False, "Task did not start as expected"
logging.debug("Will cancel transcription task")
# Stop task in progress
table_widget.selectRow(0)
# Force immediate processing of pending events before triggering cancellation
qtbot.wait(100)
window.toolbar.stop_transcription_action.trigger()
# Give some time for the cancellation to be processed
qtbot.wait(500)
qtbot.wait_until(
self._get_assert_task_status_callback(table_widget, 0, "canceled"),
timeout=60 * 1000,
)
logging.debug("Will wait for task to reach 'canceled' status")
try:
qtbot.wait_until(
self._get_assert_task_status_callback(table_widget, 0, "canceled"),
timeout=30 * 1000,
)
except Exception:
# On Windows, the cancellation might be slower, check final state
final_status = self._get_status(table_widget, 0)
logging.error(f"Task status after timeout: {final_status}")
if "canceled" not in final_status.lower():
assert False, f"Task did not cancel as expected. Final status: {final_status}"
logging.debug("Task canceled")
qtbot.wait(200)
table_widget.selectRow(0)
assert window.toolbar.stop_transcription_action.isEnabled() is False

6
tests/widgets/preferences_dialog/general_preferences_widget_test.py
View file

@ -4,7 +4,7 @@ from PyQt6.QtWidgets import QPushButton, QMessageBox, QLineEdit, QCheckBox
from buzz.locale import _
from buzz.settings.settings import Settings
from buzz.widgets.preferences_dialog.general_preferences_widget import (
GeneralPreferencesWidget, TestOpenAIApiKeyJob
GeneralPreferencesWidget, ValidateOpenAIApiKeyJob
)
@ -118,7 +118,7 @@ class TestTestOpenAIApiKeyJob:
mocker.patch('buzz.widgets.preferences_dialog.general_preferences_widget.OpenAI', return_value=mock_client)
mocker.patch('buzz.settings.settings.Settings.value', return_value="") # No custom base URL
job = TestOpenAIApiKeyJob(api_key="test_key")
job = ValidateOpenAIApiKeyJob(api_key="test_key")
mock_success = mocker.Mock()
mock_failed = mocker.Mock()
job.signals.success.connect(mock_success)
@ -140,7 +140,7 @@ class TestTestOpenAIApiKeyJob:
mocker.patch('buzz.widgets.preferences_dialog.general_preferences_widget.OpenAI', return_value=mock_client)
mocker.patch('buzz.settings.settings.Settings.value', return_value="") # No custom base URL
job = TestOpenAIApiKeyJob(api_key="wrong_key")
job = ValidateOpenAIApiKeyJob(api_key="wrong_key")
mock_success = mocker.Mock()
mock_failed = mocker.Mock()
job.signals.success.connect(mock_success)

6
tests/widgets/recording_transcriber_widget_test.py
View file

@ -111,6 +111,7 @@ class TestRecordingTranscriberWidget:
widget.close()
@pytest.mark.timeout(60)
def test_on_next_transcription_append_above(self, qtbot: QtBot):
with (patch("sounddevice.InputStream", side_effect=MockInputStream),
patch("buzz.transcriber.recording_transcriber.RecordingTranscriber.get_device_sample_rate",
@ -129,7 +130,7 @@ class TestRecordingTranscriberWidget:
widget.on_next_transcription('test2')
assert widget.transcription_text_box.toPlainText() == 'test2\n\ntest1\n\n'
qtbot.wait(200)
qtbot.wait(500)
widget.close()
@ -161,6 +162,7 @@ class TestRecordingTranscriberWidget:
assert RecordingTranscriberWidget.find_common_part("hello world", "") == ""
assert RecordingTranscriberWidget.find_common_part("", "") == ""
@pytest.mark.timeout(60)
def test_on_next_transcription_append_and_correct(self, qtbot: QtBot):
with (patch("sounddevice.InputStream", side_effect=MockInputStream),
patch("buzz.transcriber.recording_transcriber.RecordingTranscriber.get_device_sample_rate",
@ -182,6 +184,6 @@ class TestRecordingTranscriberWidget:
widget.on_next_transcription('Ceci est la deuxième phrase. Et voici la troisième.')
assert widget.transcription_text_box.toPlainText() == 'Bienvenue dans la transcription en direct de Buzz. Ceci est la deuxième phrase. Et voici la troisième.'
qtbot.wait(200)
qtbot.wait(500)
widget.close()

2
whisper.cpp

@ -1 +1 @@
Subproject commit c7b6988678779901d02ceba1a8212d2c9908956e
Subproject commit a8d002cfd879315632a579e73f0148d06959de36