Update app.py

app.py

@@ -11,7 +11,7 @@ diarization_pipe = Pipeline.from_pretrained("pyannote/speaker-diarization@2.1",
 
 
 # Load the speech-to-text model (Whisper)
-asr_pipe = pipeline("automatic-speech-recognition", model="SyedAunZaidi/whisper-small-hi")
+asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
 
 def transcribe_with_diarization(audio_path):
     # Get speaker segments using the diarization model
@@ -51,9 +51,12 @@ def transcribe_with_diarization(audio_path):
 
 iface = gr.Interface(
     fn=transcribe_with_diarization,
-    inputs=gr.Audio(source="microphone", type="filepath", filetype="wav"),
+    inputs=[
+        gr.File(label="Audio File"),
+        gr.Audio(source="microphone", type="filepath", filetype="wav")
+    ],
     outputs="text",
-    title="Whisper Large Hindi with Speaker Diarization",
+    title="Whisper small Hindi with Speaker Diarization",
     description="Real-time demo for Hindi speech recognition using a fine-tuned Whisper large model with speaker diarization.",
 )