Update app.py

app.py

@@ -9,7 +9,7 @@ from transformers import SpeechT5ForTextToSpeech, SpeechT5HifiGan, SpeechT5Proce
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
 
 # load speech translation checkpoint
-asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base", device=device,ignore_mismatched_sizes=True)
+asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-base", device=device)
 
 # load text-to-speech checkpoint and speaker embeddings
 #processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
@@ -22,7 +22,7 @@ processor = SpeechT5Processor.from_pretrained("sanchit-gandhi/speecht5_tts_vox_n
 #model = SpeechT5ForTextToSpeech.from_pretrained("jasonl1/speecht5_finetuned_voxpopuli_fi")
 model = SpeechT5ForTextToSpeech.from_pretrained("sanchit-gandhi/speecht5_tts_vox_nl",ignore_mismatched_sizes=True)
 
-vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan").to(device)
+vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan",ignore_mismatched_sizes=True).to(device)
 
 embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
 speaker_embeddings = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0)