Zengwei Yao

Thanks for your contribution! I have some questions. It looks like the streaming zipformer code `zipformer.py` is copied from pruned7_steaming recipe, which should be a soft link instead. I am...

Which decoding script did you use, `decode.py` or `streaming_decode.py`? I think the issue might be caused by the tail padding. You could try to reduce the tail padding length: https://github.com/k2-fsa/icefall/blob/b87ed26c09e9f5bb29174dd01f13670fb6124583/egs/librispeech/ASR/zipformer/decode.py#L439...

You could also try setting `length_norm` to `False` for `modified_beam_search` when using `streaming_decode.py`, (just for debugging) https://github.com/k2-fsa/icefall/blob/b87ed26c09e9f5bb29174dd01f13670fb6124583/egs/librispeech/ASR/zipformer/decode_stream.py#L144

We did try using context-size of 3 or 4 (at least in Zipformer), but could not get improvements.

@AdolfVonKleist Did you get difference between `decode.py` and `streaming_decode.py` for a same audio?

I suggest to use the lastest recipe `zipformer` instead. In the old recipe `pruned_transducer_stateless7_streaming`, there might be some issues when doing the chunk-wise forward for the first chunks, since we...

> > > > I suggest to use the lastest recipe `zipformer` instead. In the old recipe `pruned_transducer_stateless7_streaming`, there might be some issues when doing the chunk-wise forward for the...

> A minor comment: In Figure 3, the color of the dashed lines representing the E-branchformer and conformer is not easily distinguishable for people with color blindness. Thanks for your...

> could you also update https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/RESULTS-100hours.md OK. I will do that.

Sure. I will have a look.