Update README.

This commit is contained in:
Fangjun Kuang 2021-11-09 20:55:00 +08:00
parent 1e4920410f
commit 86604b197d
2 changed files with 12 additions and 7 deletions

View File

@ -39,9 +39,10 @@ and [TDNN LSTM CTC model][LibriSpeech_tdnn_lstm_ctc].
The best WER we currently have is: The best WER we currently have is:
||test-clean|test-other| | | test-clean | test-other |
|--|--|--| |-----|------------|------------|
|WER| 2.57% | 5.94% | | WER | 2.42 | 5.73 |
We provide a Colab notebook to run a pre-trained conformer CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1huyupXAcHsUrKaWfI83iMEJ6J0Nh0213?usp=sharing) We provide a Colab notebook to run a pre-trained conformer CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1huyupXAcHsUrKaWfI83iMEJ6J0Nh0213?usp=sharing)
@ -49,9 +50,9 @@ We provide a Colab notebook to run a pre-trained conformer CTC model: [![Open In
The WER for this model is: The WER for this model is:
||test-clean|test-other| | | test-clean | test-other |
|--|--|--| |-----|------------|------------|
|WER| 6.59% | 17.69% | | WER | 6.59 | 17.69 |
We provide a Colab notebook to run a pre-trained TDNN LSTM CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1kNmDXNMwREi0rZGAOIAOJo93REBuOTcd?usp=sharing) We provide a Colab notebook to run a pre-trained TDNN LSTM CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1kNmDXNMwREi0rZGAOIAOJo93REBuOTcd?usp=sharing)

View File

@ -33,7 +33,8 @@ export CUDA_VISIBLE_DEVICES="0,1,2,3"
--world-size 4 \ --world-size 4 \
--bucketing-sampler 1 \ --bucketing-sampler 1 \
--start-epoch 0 \ --start-epoch 0 \
--num-epochs 80 --num-epochs 90
# Note: It trains for 90 epochs, but the best WER is at epoch-77.pt
``` ```
and the following command for decoding and the following command for decoding
@ -55,6 +56,9 @@ and the following command for decoding
You can find the pre-trained model by visiting You can find the pre-trained model by visiting
<https://huggingface.co/csukuangfj/icefall-asr-librispeech-conformer-ctc-jit-bpe-500-2021-11-09> <https://huggingface.co/csukuangfj/icefall-asr-librispeech-conformer-ctc-jit-bpe-500-2021-11-09>
The tensorboard log for training is available at
<https://tensorboard.dev/experiment/hZDWrZfaSqOMqtW0NEfXKg/#scalars>
#### 2021-08-19 #### 2021-08-19
(Wei Kang): Result of https://github.com/k2-fsa/icefall/pull/13 (Wei Kang): Result of https://github.com/k2-fsa/icefall/pull/13