asr

app.py

@@ -8,7 +8,7 @@ mms_transcribe = gr.Interface(
     inputs=[
         gr.Audio(),
         gr.Dropdown(
-            choices=[model for model in asr.models_info],
+            choices=[model for model in asr.models_info] + ["Compare All Models"],
             label="Select Model for ASR",
             value="ixxan/wav2vec2-large-mms-1b-uyghur-latin",
             interactive=True

asr.py

@@ -40,14 +40,18 @@ models_info = {
 }
 
 def transcribe(audio_data, model_id) -> str:
-    """
-    Transcribes audio to text using the Whisper model for Uyghur.
-    Args:
-    - audio_data: Gradio audio input 
-    Returns:
-    - str: The transcription of the audio.
-    """
+    if model_id == "Compare All Models":
+        return transcribe_all_models(audio_data)
+    else:
+        return transcribe_with_model(audio_data, model_id)
+
+def transcribe_all_models(audio_data) -> dict:
+    transcriptions = {}
+    for model_id in models_info.keys():
+        transcriptions[model_id] = transcribe_with_model(audio_data, model_id)
+    return transcriptions
 
+def transcribe_with_model(audio_data, model_id) -> str:
     # Load audio file
     if not audio_data:
         return "<<ERROR: Empty Audio Input>>"