icefall

icefall copied to clipboard

Get n best results

1

Is there any way I can get n best decoding results for rescoring?

ncakhoa

on-the-fly fbank feats

17

Hey guys, I notice there‘s on-the-fly feats in `asr_datamodule.py`: https://github.com/k2-fsa/icefall/blob/32de2766d591d2e1a77c06a40d2861fb1bbcd3ad/egs/wenetspeech/ASR/pruned_transducer_stateless2/asr_datamodule.py#L279-L298 However, I didn't find any recipe using that feats, how could I using on-the-fly feats instead of making fbank first(I'm...

Cescfangs

streaming_decode.py or decode.py under stateless5?

2

Hey guys, I see both `streaming_decode.py` and `decode.py` under stateless5 support streaming decoding, what's the difference?

Cescfangs

inf pruned_loss and high simple_loss in middle of training

21

I'm training an asr model using conv_emformer_transducer_stateless2 with my own data. In the middle of training I get the following error: **simple_loss: 169.94381713867188 pruned_loss: inf** Here are some info about...

is2022

WIP: new recipe: lightweight CTC model for librispeech

4

This PR includes: 1. A new recipe with lightweight CTC model. The encoder is similar to MobileNet V2 with a param count of just 1.76M using phone Lexicon and default...

wangtiance

[Streaming Conformer] Causal ConvolutionModule -> streaming ONNX/torch results mismatch

2

Hi, I've tried to convert [Conformer encoder](https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/pruned_transducer_stateless2/conformer.py) for streaming purposes to ONNX using parts of sherpa's script https://github.com/k2-fsa/sherpa/blob/master/triton/scripts/export_onnx.py. If I set the model `causal=False` the mean difference between torch and...

MicKot

This PR contains some changes we did during our icassp paper preparation; it also supports LG decoding for Chinese dataset (Aishell, WenetSpeech).

pkufool

The difference between forward score and backward score exceeds 1.0

2

Thank you always. It's been very helpful since kaldi : ) I am trying to train `[librispeech/ASR/conformer_mmi](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/conformer_mmi) with a private dataset to check the recipe works well. I checked that...

somniumism

Hi, I've been facing this flake8 error for pre-commit, and this was the fix I found to be added in pre-commit-config.yaml. ```yaml - repo: https://github.com/PyCQA/flake8 rev: 3.9.2 hooks: - id:...

teowenshen

How to test audio files with model trained on Gigaspeech

1

Hi authors, Do you have any suggestions about how to test audio files with model trained on Gigaspeech (you provided)? Thanks so much.

WangHelin1997