Update app.py

app.py

@@ -15,7 +15,6 @@ speaker_model = EncDecSpeakerLabelModel.from_pretrained("nvidia/speakerverificat
 model.eval()
 
 def run_diarization(path1):
-    print(path1)
     annotation = model(path1, num_workers=0, batch_size=16)
     rttm=annotation.to_rttm()
     df = pd.DataFrame(columns=['start_time', 'end_time', 'speaker', 'text'])
@@ -65,7 +64,7 @@ def get_transcripts(df, audio_path):
     filename = create_manifest(df,audio_path)
     model = EncDecRNNTBPEModel.from_pretrained(model_name="nvidia/stt_en_fastconformer_transducer_large").to(device)
     model.eval()
-    config = OmegaConf.create({"manifest_filepath": filename, 'batch_size': 4})
+    config = OmegaConf.create({"manifest_filepath": filename, 'batch_size': 2})
     dataloader = model._setup_transcribe_dataloader(config)
     
     hypotheses = []
@@ -106,8 +105,9 @@ examples = [
 
 microphone_interface = gr.Interface(
     fn=run_diarization,
-    inputs=[gr.Audio(source="microphone", type="filepath", optional=True, label="Mic Audio")],
-    outputs=[gr.components.Dataframe()],
+    inputs=[gr.Audio(source="microphone", type="filepath", label="Mic Audio")],
+    outputs=[gr.components.Dataframe(wrap=True, label='Speaker Diariazation with Speech Recognition',
+        row_count=(1, "dynamic"), headers=['start_time', 'end_time', 'speaker', 'text'])],
     title="Offline Speaker Diarization with NeMo",
     description="This demonstration will perform offline speaker diarization on an audio file using nemo",
     article=article,
@@ -116,12 +116,13 @@ microphone_interface = gr.Interface(
     allow_flagging=False,
     live=False,
     examples=examples,
-)
+    )   
 
 upload_interface = gr.Interface(
     fn=run_diarization,
-    inputs=[gr.Audio(source="upload", type='filepath',optional=True, label='Upload File')],
-    outputs=[gr.components.Dataframe()],
+    inputs=[gr.Audio(source="upload", type='filepath', label='Upload File')],
+    outputs=[gr.components.Dataframe(wrap=True, label='Speaker Diariazation with Speech Recognition',
+        row_count=(1, "dynamic"), headers=['start_time', 'end_time', 'speaker', 'text'])],
     title="Offline Speaker Diarization with NeMo",
     description="This demonstration will perform offline speaker diarization on an audio file using nemo",
     article=article,
@@ -130,8 +131,8 @@ upload_interface = gr.Interface(
     allow_flagging=False,
     live=False,
     examples=examples,
-)
+    )
 
 demo = gr.TabbedInterface([microphone_interface, upload_interface], ["Microphone", "Upload File"])
 
-demo.launch(enable_queue=True)
+demo.launch(enable_queue=True)