Update app.py

app.py

@@ -3,10 +3,22 @@ import numpy as np
 import torch
 from transformers import pipeline, VitsModel, AutoTokenizer, AutoTokenizer
 from transformers import SpeechT5ForTextToSpeech, SpeechT5HifiGan, SpeechT5Processor
+from transformers import WhisperTokenizer, GenerationConfig
+
 
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
 
-asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-large-v2", device=device)
+
+tokenizer = WhisperTokenizer.from_pretrained("openai/whisper-base")
+generation_config = GenerationConfig.from_pretrained("openai/whisper-base")
+
+generation_config.forced_decoder_ids
+
+tokenizer.decode(generation_config.forced_decoder_ids[1][1])
+
+
+
+asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-medium", device=device)
 
 #vist_model = VitsModel.from_pretrained("facebook/mms-tts-spa")
 #vist_tokenizer = AutoTokenizer.from_pretrained("facebook/mms-tts-spa")