Update app.py

app.py

@@ -22,15 +22,7 @@ tts = TextToSpeech(kv_cache=True, use_deepspeed=True, half=True)
 
 def convert_audio(filepath, voice="indian_F_1"):
     # Transcribe audio to text using STT
-    transcription_output = pipe(
-        filepath,
-        max_new_tokens=256,
-        generate_kwargs={
-            "task": "transcribe",
-            "language": "english",
-        },
-        chunk_length_s=30,
-        batch_size=8
+    
     )
     transcribed_text = transcription_output["text"]
     
@@ -42,15 +34,14 @@ def convert_audio(filepath, voice="indian_F_1"):
         for audio_frame in tts.tts_with_preset(
             text,
             voice_samples=voice_samples,
-            conditioning_latents=conditioning_latents,
-            preset="ultra_fast",
+            
             k=1
         ):
             audio_frames.append(audio_frame.cpu().detach().numpy())
 
     # Joining the audio frames for output using numpy's concatenate
     final_audio = np.concatenate(audio_frames, axis=0)
-    return (24000, final_audio)
+    
 
 interface = gr.Interface(
     fn=convert_audio,