Update app.py

app.py

@@ -1,4 +1,4 @@
-from turtle import title
+#from turtle import title
 import gradio as gr
 
 import git
@@ -7,6 +7,11 @@ os.system('git clone https://github.com/Edresson/Coqui-TTS -b multilingual-torch
 os.system('pip install -q -e TTS/')
 os.system('pip install -q torchaudio==0.9.0')
 
+os.system('pip install voicefixer --upgrade')
+from voicefixer import VoiceFixer
+voicefixer = VoiceFixer()
+
+
 import sys
 TTS_PATH = "TTS/"
 
@@ -23,9 +28,14 @@ import numpy as np
 import IPython
 from IPython.display import Audio
 
-
 import torch
 
+enhance_model = SpectralMaskEnhancement.from_hparams(
+source="speechbrain/metricgan-plus-voicebank",
+savedir="pretrained_models/metricgan-plus-voicebank",
+run_opts={"device":"cuda"},
+)
+
 from TTS.tts.utils.synthesis import synthesis
 from TTS.tts.utils.text.symbols import make_symbols, phonemes, symbols
 try:
@@ -154,12 +164,27 @@ def greet(Text,Voicetoclone,VoiceMicrophone):
       out_path = os.path.join(OUT_PATH, file_name)
       print(" > Saving output to {}".format(out_path))
       ap.save_wav(wav, out_path)
-      return 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, or 2 to find out the best result
+    
+      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)
+
+      return "enhanced.wav"
 
 demo = gr.Interface(
     fn=greet, 
-    inputs=[gr.inputs.Textbox(label='What would you like the voice to say? (max. 2000 characters per request)'),gr.Audio(type="filepath",         source="upload",label='Please upload a voice to clone (max. 30mb)'),gr.Audio(source="microphone", type="filepath", streaming=True)],
+    inputs=[gr.inputs.Textbox(label='请输入您想要合成的文字，请自觉合法合规使用！'),gr.Audio(type="filepath", source="upload",label='请上传您喜欢的声音(wav/mp3文件, max. 30mb)'),gr.Audio(source="microphone", type="filepath", label = '请用麦克风上传您喜欢的声音，与文件上传二选一即可')],
     outputs="audio",
-    title="Bilal's Voice Cloning Tool"
+    title="🥳💬💕 - TalktoAI，随时随地，谈天说地！",
+    description = "🔴注意🔴：请不要生成会对个人以及组织造成侵害的内容，仅供科研、学习使用。用户生成内容与程序开发者无关，请自觉合法合规使用，违反者一切后果自负。"
+    article = "🤖 - 让有人文关怀的AI造福每一个人！AI向善，文明璀璨！TalktoAI - Enable the future！",
     )
 demo.launch()