juancopi81 commited on
Commit
cb8134f
Β·
1 Parent(s): 8ff7dd3

Update app.py

Browse files

Change to spanish

Files changed (1) hide show
  1. app.py +2 -2
app.py CHANGED
@@ -14,7 +14,7 @@ asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base",
14
  # load text-to-speech checkpoint and speaker embeddings
15
  processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
16
 
17
- model = SpeechT5ForTextToSpeech.from_pretrained("microsoft/speecht5_tts").to(device)
18
  vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to(device)
19
 
20
  embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
@@ -22,7 +22,7 @@ speaker_embeddings = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze
22
 
23
 
24
  def translate(audio):
25
- outputs = asr_pipe(audio, max_new_tokens=256, generate_kwargs={"task": "translate"})
26
  return outputs["text"]
27
 
28
 
 
14
  # load text-to-speech checkpoint and speaker embeddings
15
  processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
16
 
17
+ model = SpeechT5ForTextToSpeech.from_pretrained("juancopi81/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")
 
22
 
23
 
24
  def translate(audio):
25
+ outputs = asr_pipe(audio, max_new_tokens=256, generate_kwargs={"task": "translate", "language": "es"})
26
  return outputs["text"]
27
 
28