Update app.py

app.py

@@ -6,21 +6,22 @@ def load_model(model_size):
     model = WhisperModel(model_size, device="cpu", compute_type="int8")
     return model
 
-# Transcription function
+# Streaming transcription function
 def transcribe_audio(model_size, audio_file):
     # Initialize the model with the given size
     model = load_model(model_size)
     
-    # Transcribe the audio file
+    # Stream the transcription of the audio file
+    transcribed_text = ""
     segments, info = model.transcribe(audio_file, beam_size=5)
+
+    # Yield detected language information first
+    yield f"Detected language: {info.language} (Probability: {info.language_probability:.2f})", transcribed_text.strip()
     
-    # Extract the transcribed text
-    transcribed_text = ""
+    # Then yield each segment of transcribed text as it is processed
     for segment in segments:
         transcribed_text += segment.text + " "
-
-    # Return detected language and transcribed text
-    return f"Detected language: {info.language} (Probability: {info.language_probability:.2f})", transcribed_text.strip()
+        yield "", transcribed_text.strip()  # Empty string for language, we only update transcription
 
 # Define the Gradio interface
 interface = gr.Interface(
@@ -34,7 +35,8 @@ interface = gr.Interface(
         gr.Textbox(label="Transcription")
     ],  # Output language and transcription
     title="Whisper Transcription App",
-    description="Upload an audio file and specify the model size to transcribe it using WhisperModel."
+    description="Upload an audio file and specify the model size to transcribe it using WhisperModel.",
+    live=True  # Enable live updates
 )
 
 # Launch the app