Update app.py

app.py

@@ -27,7 +27,7 @@ model_name = "pgilles/whisper-large-v2-lb_cased_01"
 #model = model.to(device)
 
 #p = pipeline("automatic-speech-recognition", model=model, tokenizer=tokenizer, feature_extractor=processor.feature_extractor, decoder=processor.decoder, use_auth_token=token_key)
-p = pipeline("automatic-speech-recognition", model=model_name, device=0)
+p = pipeline("automatic-speech-recognition", model=model_name, chunk_length_s = 29, device=0)
 
 #p = pipeline("automatic-speech-recognition", model=model_name, use_auth_token = token_key)
 #p = pipeline("automatic-speech-recognition", model=model_name, use_auth_token = True)
@@ -49,9 +49,14 @@ def asr_pipe(input_file):
   load_data(input_file)
   transcription = p(input_file, chunk_length_s=3, stride_length_s=(0.5, 0.5))["text"]
   return transcription
-    
+
+input1=[gr.inputs.Audio(source="microphone", type='filepath', label="Click and Start Speaking..."), "state"]
+input2=gr.inputs.Audio(source="upload", type='filepath', label="Load your own file...")
+inputs=[input1, input2]
+
 gr.Interface(asr_pipe,
-             inputs = gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Hei kënnt Dir Är Sprooch iwwert de Mikro ophuelen"),
+             #inputs = gr.inputs.Audio(source="microphone", type="filepath", optional=True, label="Hei kënnt Dir Är Sprooch iwwert de Mikro ophuelen"),
+             inputs = inputs,
              outputs = gr.outputs.Textbox(label="Erkannten Text"),
              title="Sproocherkennung fir d'Lëtzebuergescht @uni.lu",
              description = "Dës App convertéiert Är geschwate Sprooch an de (méi oder manner richtegen ;-)) Text! Dir kënnt Iech selwer iwwer de Mikro ophuelen (am beschten 5 bis 10 Sekonnen) oder e Beispill ënnen auswielen. Den erkannten Text gëtt da riets gewisen. D''Word Error Rate' vun dësem Modell läit bei 14, d.h. vun 100 Wierder ginn der 14 net richteg erkannt.",