Update app.py

app.py

@@ -40,13 +40,13 @@ from TTS.tts.models.vits import *
 
 import whisper
 model = whisper.load_model("base")
-import os
+
 os.system('pip install voicefixer --upgrade')
 from voicefixer import VoiceFixer
 voicefixer = VoiceFixer()
-import gradio as gr
+
 import openai
-import torch
+
 import torchaudio
 from speechbrain.pretrained import SpectralMaskEnhancement
 
@@ -135,7 +135,7 @@ def compute_spec(ref_file):
   
 
     
-def greet(apikey,Voicetoclone,VoiceMicrophone, audio, choice1):
+def greet(apikey, Voicetoclone, VoiceMicrophone, audio, choice1):
 
     openai.api_key = apikey
 
@@ -175,7 +175,7 @@ def greet(apikey,Voicetoclone,VoiceMicrophone, audio, choice1):
 
     messages.append({"role": "assistant", "content": chat_response})   
     
-    text = chat_response
+    text= "%s" % (chat_response)
     if Voicetoclone is not None:
       reference_files= "%s" % (Voicetoclone)
       print("path url")
@@ -226,21 +226,21 @@ def greet(apikey,Voicetoclone,VoiceMicrophone, audio, choice1):
       print(" > Saving output to {}".format(out_path))
       ap.save_wav(wav, out_path)
 
-#      voicefixer.restore(input=out_path, # input wav file path
- #                     output="audio1.wav", # output wav file path
-  #                    cuda=True, # whether to use gpu acceleration
-   #                   mode = 0) # You can try out mode 0, 1 to find out the best result
+      voicefixer.restore(input=out_path, # input wav file path
+                      output="audio1.wav", # output wav file path
+                      cuda=True, # whether to use gpu acceleration
+                      mode = 0) # You can try out mode 0, 1 to find out the best result
 
 
 
-#      noisy = enhance_model.load_audio(
- #     "audio1.wav"
-  #    ).unsqueeze(0)
+      noisy = enhance_model.load_audio(
+      "audio1.wav"
+      ).unsqueeze(0)
 
-#      enhanced = enhance_model.enhance_batch(noisy, lengths=torch.tensor([1.]))
-#      torchaudio.save("enhanced.wav", enhanced.cpu(), 16000)
+      enhanced = enhance_model.enhance_batch(noisy, lengths=torch.tensor([1.]))
+      torchaudio.save("enhanced.wav", enhanced.cpu(), 16000)
 
-      return [result.text, chat_response, out_path]
+      return [result.text, chat_response, "enhanced.wav"]
 
 output_1 = gr.Textbox(label="Speech to Text")
 output_2 = gr.Textbox(label="ChatGPT Output")