语言: 英语
数据集:

• librispeech

标签:

• 音频
• 自动语音识别
• 语音
• asr
• hubert

许可证: apache-2.0

评估指标:

• wer
• cer

voidful/tts_hubert_cluster_bart_base

使用方法

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM  
tokenizer = AutoTokenizer.from_pretrained("voidful/tts_hubert_cluster_bart_base")  
model = AutoModelForSeq2SeqLM.from_pretrained("voidful/tts_hubert_cluster_bart_base")  

生成输出

gen_output = model.generate(input_ids=tokenizer("沿着泥泞的乡间小路前行，一连两周在潮湿的校舍里对着湿漉漉的听众演讲，他必须在周日早晨出现在某个礼拜场所露个面，之后就能立即前来询问", return_tensors='pt').input_ids, max_length=1024)  
print(tokenizer.decode(gen_output[0], skip_special_tokens=True))  

结果

:vtok402::vtok329::vtok329::vtok75::vtok75::vtok75::vtok44::vtok150::vtok150::vtok222::vtok280::vtok280::vtok138::vtok409::vtok409::vtok409::vtok46::vtok441: