zhifu gao
zhifu gao
If it is convenient to you, please upload the test audio and we could verify it.
Please raise the issue as https://github.com/alibaba-damo-academy/FunASR/issues/1074
Sorry, it is a bug. We will fix it by 20, Sept.
We have fixed this bug. Please update modelscope-1.9.2 `pip install -U modescope>=1.9.2`
中文模板，请参考：https://github.com/alibaba-damo-academy/FunASR/issues/1074
（建议英文提问，不强制） OS: [e.g. linux] Python/C++ Version： Package Version：pytorch、torchaudio、modelscope、funasr version（pip list） Model： Command： Details： Error log：
The max duration of UniASR is 15s. If your wav is longer, you should use a vad to split the wav into short slices. https://github.com/alibaba-damo-academy/FunASR/discussions/278
The `apk` could be found in https://www.modelscope.cn/models/manyeyes/aliparaformerasr-large-model-offline/summary The source is in https://github.com/alibaba-damo-academy/FunASR/blob/main/runtime/csharp/AliParaformerAsr/README.md
Maybe you could change the code and try it again, in `egs/aishell/paraformer/utils/apply_cmvn.sh` line 20: ```shell #$cmd JOB=1:$nj $logdir/apply_cmvn.JOB.log \ # python utils/apply_cmvn.py -a $fbankdir/ark/feats.JOB.ark \ # -c $cmvn_file -i JOB...
For the gpu usage, you should export torchscripts on gpu and keep the envs same.