TianHao Zhang

[s2t] wav2vec2 pipeline

2

### PR types New features ### PR changes Models ### Describe wav2vec2 pipeline

[Discussion] How to use spleeter with multi gpus?

3

I want to use spleeter to process a large amount of video data, and I want to use multi gpus to speed up my processing, is multi gpus currently supported?...

【BUG】 ModuleNotFoundError: No module named 'fun_text_processing'

2

when i use itn model, an error in here: https://github.com/modelscope/modelscope/blob/fb46e18351be42927042bc06b1aef80450b2062c/modelscope/pipelines/audio/inverse_text_processing_pipeline.py#L106 how can i import the fun_text_processing module ?

[Audio LLM]support audiollm for asr, based on whisper and llama3

6

conduct experiment on librispeech dataset for severl steps:

请问是否可以支持流式dataset？ 当前对large-scale data处理的时候耗时非常长，并且会缓存非常大的cache，请问目前有没有简便的办法改为iterabledataset，或者有没有其他推荐的对large-scale data的处理方式？

流式推理存在音色变化问题

6

在流式推理的过程中，发现这样的badcase： - 合成语音中从第二句话开始音色发生了变化，似乎非流式没有遇到这个问题。 请问有没有遇到过类似的问题？大概的原因和出现概率是咋样的？

url_iwslt19 is not valid in the how2 example

4

When I run run.sh under example2/how，an error occurs:  this url is valid and I find in #4866 has propose this bug， but it seems that not fix in #4899...