README.md

metadata

tags:
  - espnet
  - audio
  - speech-recognition
language: en
datasets:
  - google/fleurs
license: cc-by-4.0

ESPnet2 ASR model

espnet/wanchichen_fleurs_english_asr_wav2vec_frontend

This model was trained by William Chen using the fleurs recipe in espnet.

Demo: How to use in ESPnet2

cd espnet
pip install -e .
cd egs2/fleurs/asr1
./run.sh

RESULTS

Environments

• date: Sun Aug 14 14:52:04 EDT 2022
• python version: 3.8.6 (default, Dec 17 2020, 16:57:01) [GCC 10.2.0]
• espnet version: espnet 202205
• pytorch version: pytorch 1.8.1+cu102
• Git hash: 45e8cb9173a072f85ee7a7ccbcae06af7c5c484a
  • Commit date: Wed Jun 1 14:21:14 2022 +0900

asr_train_asr_wav2vec_960h_transformer_raw_en_us_bpe300_sp

WER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decode_asr_asr_model_valid.acc.best/test_all	647	14344	67.1	29.4	3.5	4.6	37.5	99.8
decode_asr_asr_model_valid.acc.best/dev_all	388	7935	66.8	29.7	3.6	5.0	38.2	99.0

CER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decode_asr_asr_model_valid.acc.best/test_all	647	83954	88.6	5.4	6.0	4.8	16.2	99.8
decode_asr_asr_model_valid.acc.best/dev_all	388	47051	88.1	6.0	5.9	4.4	16.3	99.0

TER

dataset	Snt	Wrd	Corr	Sub	Del	Ins	Err	S.Err
decode_asr_asr_model_valid.acc.best/test_all	647	39965	7.7	14.9	7.4	4.1	26.4	99.8
decode_asr_asr_model_valid.acc.best/dev_all	388	22491	77.3	15.2	7.5	3.8	26.5	99.0