Update app.py

app.py

@@ -1,8 +1,15 @@
+print("import gradio")
 import gradio as gr
+print("import ppn")
 from pypinyin import lazy_pinyin, Style
+print("import torch")
 import torch
+device = 'cuda:0' if torch.cuda.is_available() else 'cpu'
+print("import ttts")
 from ttts.utils.infer_utils import load_model
+print("import mel")
 from ttts.vocoder.feature_extractors import MelSpectrogramFeatures
+print("import torchaudio")
 import torchaudio
 MODELS = {
     'vqvae.pth':'./TTTS/vae-30.pt',
@@ -13,9 +20,12 @@ MODELS = {
     'rlg_auto.pth': '',
     'rlg_diffuser.pth': '',
 }
+print("import tokenizer")
 from ttts.gpt.voice_tokenizer import VoiceBpeTokenizer
+print("import f")
 import torch.nn.functional as F
 cond_audio = 'ttts/3.wav'
+print("load audio")
 audio,sr = torchaudio.load(cond_audio)
 if audio.shape[0]>1:
     audio = audio[0].unsqueeze(0)
@@ -39,7 +49,7 @@ from ttts.diffusion.aa_model import denormalize_tacotron_mel, normalize_tacotron
 # print(device)
 
 vocos = Vocos.from_pretrained("charactr/vocos-mel-24khz")
-device = 'cuda:0' if torch.cuda.is_available() else 'cpu'
+
 
 def speak(text):
     pinyin = ' '.join(lazy_pinyin(text, style=Style.TONE3, neutral_tone_with_five=True))