Update app.py

app.py

@@ -1,98 +1,64 @@
 from turtle import title
 import gradio as gr
-
 import git
 import os
 os.system('git clone https://github.com/Edresson/Coqui-TTS -b multilingual-torchaudio-SE TTS')
 os.system('pip install -q -e TTS/')
 os.system('pip install -q torchaudio==0.9.0')
-
 import sys
 TTS_PATH = "TTS/"
-
-# add libraries into environment
 sys.path.append(TTS_PATH) # set this if TTS is not installed globally
-
 import os
 import string
 import time
 import argparse
 import json
-
 import numpy as np
 import IPython
 from IPython.display import Audio
-
-
 import torch
-
 from TTS.tts.utils.synthesis import synthesis
 from TTS.tts.utils.text.symbols import make_symbols, phonemes, symbols
 try:
   from TTS.utils.audio import AudioProcessor
 except:
   from TTS.utils.audio import AudioProcessor
-
-
 from TTS.tts.models import setup_model
 from TTS.config import load_config
 from TTS.tts.models.vits import *  
 
 OUT_PATH = 'out/'
-
-# create output path
 os.makedirs(OUT_PATH, exist_ok=True)
 
-# model vars 
 MODEL_PATH = '/home/user/app/best_model_latest.pth.tar'
 CONFIG_PATH = '/home/user/app/config.json'
 TTS_LANGUAGES = "/home/user/app/language_ids.json"
 TTS_SPEAKERS = "/home/user/app/speakers.json"
 USE_CUDA = torch.cuda.is_available()  
 
-# load the config
 C = load_config(CONFIG_PATH)
-
-
-# load the audio processor
 ap = AudioProcessor(**C.audio)
-
 speaker_embedding = None
-
 C.model_args['d_vector_file'] = TTS_SPEAKERS
 C.model_args['use_speaker_encoder_as_loss'] = False
-
 model = setup_model(C)
 model.language_manager.set_language_ids_from_file(TTS_LANGUAGES)
-# print(model.language_manager.num_languages, model.embedded_language_dim)
-# print(model.emb_l)
 cp = torch.load(MODEL_PATH, map_location=torch.device('cpu'))
-# remove speaker encoder
 model_weights = cp['model'].copy()
 for key in list(model_weights.keys()):
   if "speaker_encoder" in key:
     del model_weights[key]
-
 model.load_state_dict(model_weights)
-
-
 model.eval()
-
 if USE_CUDA:
     model = model.cuda()
-
-# synthesize voice
 use_griffin_lim = False
-
 os.system('pip install -q pydub ffmpeg-normalize')
-
 CONFIG_SE_PATH = "config_se.json"
 CHECKPOINT_SE_PATH = "SE_checkpoint.pth.tar"
-
 from TTS.tts.utils.speakers import SpeakerManager
 from pydub import AudioSegment
 import librosa
-
 SE_speaker_manager = SpeakerManager(encoder_model_path=CHECKPOINT_SE_PATH, encoder_config_path=CONFIG_SE_PATH, use_cuda=USE_CUDA)
 
 def compute_spec(ref_file):
@@ -101,8 +67,6 @@ def compute_spec(ref_file):
   spec = torch.FloatTensor(spec).unsqueeze(0)
   return spec
   
-
-    
 def greet(Text,Voicetoclone,VoiceMicrophone):
     text= "%s" % (Text)
     if Voicetoclone is not None:
@@ -130,7 +94,6 @@ def greet(Text,Voicetoclone,VoiceMicrophone):
       text = text
       model.language_manager.language_id_mapping
       language_id = 0
-    
       print(" > text: {}".format(text))
       wav, alignment, _, _ = synthesis(
                         model,
@@ -160,6 +123,6 @@ 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)],
     outputs="audio",
-    title="Bilal's Voice Cloning Tool"
+    title="Clone Any Voice"
     )
 demo.launch()