diff --git a/buzz/widgets/transcription_viewer/speaker_identification_widget.py b/buzz/widgets/transcription_viewer/speaker_identification_widget.py
index cc794419..67408256 100644
--- a/buzz/widgets/transcription_viewer/speaker_identification_widget.py
+++ b/buzz/widgets/transcription_viewer/speaker_identification_widget.py
@@ -187,7 +187,8 @@ class IdentificationWorker(QObject):
                 transcription_id=self.transcription.id_as_uuid
             )
 
-            full_transcript = "".join(segment.text for segment in segments)
+            full_transcript = " ".join(segment.text for segment in segments)
+            full_transcript = re.sub(r' {2,}', ' ', full_transcript)
 
             if self._is_cancelled:
                 logging.debug("Speaker identification worker: Cancelled at step 2")
diff --git a/tests/widgets/speaker_identification_widget_test.py b/tests/widgets/speaker_identification_widget_test.py
index 5f10e6ce..54dc4071 100644
--- a/tests/widgets/speaker_identification_widget_test.py
+++ b/tests/widgets/speaker_identification_widget_test.py
@@ -87,7 +87,8 @@ class TestSpeakerIdentificationWidget:
         assert worker.transcription == transcription
         assert len(result) == 1
         assert isinstance(result[0], list)
-        assert result == [[{'end_time': 8904, 'speaker': 'Speaker 0', 'start_time': 140, 'text': 'Bienvenue dans. '}]]
+        assert (result == [[{'end_time': 8904, 'speaker': 'Speaker 0', 'start_time': 140, 'text': 'Bien venue dans. '}]]
+                or result == [[{'end_time': 8904, 'speaker': 'Speaker 0', 'start_time': 140, 'text': 'Bienvenue dans. '}]])
 
     def test_batch_processing_with_many_words(self):
         """Test batch processing when there are more than 200 words."""