Dario Wisznewer commited on
Commit
56a4ec7
·
unverified ·
1 Parent(s): 91a782e

Updated whisper model to a bigger one, changed tts checkpoint and updated comment

Browse files
Files changed (1) hide show
  1. app.py +3 -3
app.py CHANGED
@@ -9,12 +9,12 @@ from transformers import SpeechT5ForTextToSpeech, SpeechT5HifiGan, SpeechT5Proce
9
  device = "cuda:0" if torch.cuda.is_available() else "cpu"
10
 
11
  # load speech translation checkpoint
12
- asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base", device=device)
13
 
14
  # load text-to-speech checkpoint and speaker embeddings
15
- processor = SpeechT5Processor.from_pretrained("ptah23/speecht5_finetuned_voxpopuli_es")
16
 
17
- model = SpeechT5ForTextToSpeech.from_pretrained("ptah23/speecht5_finetuned_voxpopuli_es").to(device)
18
  vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to(device)
19
 
20
  embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
 
9
  device = "cuda:0" if torch.cuda.is_available() else "cpu"
10
 
11
  # load speech translation checkpoint
12
+ asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-large-v2", device=device)
13
 
14
  # load text-to-speech checkpoint and speaker embeddings
15
+ processor = SpeechT5Processor.from_pretrained("Sandiago21/speecht5_finetuned_facebook_voxpopuli_spanish")
16
 
17
+ model = SpeechT5ForTextToSpeech.from_pretrained("Sandiago21/speecht5_finetuned_facebook_voxpopuli_spanish").to(device)
18
  vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to(device)
19
 
20
  embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")