mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-12-11 06:55:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Minor fixes.

Browse Source
This commit is contained in:
Fangjun Kuang 2022-02-21 15:11:46 +08:00
parent 9f69dafc92
commit aadd7cae42
3 changed files with 38 additions and 33 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
egs/librispeech/ASR/transducer_stateless_multi_datasets/README.md
View File

@ -8,15 +8,20 @@ You can use the following command to start the training:
```bash
cd egs/librispeech/ASR
./prepare.sh
./prepare_giga_speech.sh
export CUDA_VISIBLE_DEVICES="0,1,2,3"
export CUDA_VISIBLE_DEVICES="0,1"
./transducer_stateless/train.py \
--world-size 4 \
--num-epochs 30 \
./transducer_stateless_multi_datasets/train.py \
--world-size 2 \
--num-epochs 60 \
--start-epoch 0 \
--exp-dir transducer_stateless/exp \
--full-libri 1 \
--max-duration 250 \
--lr-factor 2.5
--exp-dir transducer_stateless_multi_datasets/exp-100 \
--full-libri 0 \
--max-duration 300 \
--lr-factor 1 \
--bpe-model data/lang_bpe_500/bpe.model \
--modified-transducer-prob 0.25
--giga-prob 0.2
```

48
egs/librispeech/ASR/transducer_stateless_multi_datasets/pretrained.py
View File

@ -18,36 +18,36 @@
Usage:
(1) greedy search
./transducer_stateless/pretrained.py \
--checkpoint ./transducer_stateless/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method greedy_search \
--max-sym-per-frame 1 \
/path/to/foo.wav \
/path/to/bar.wav \
./transducer_stateless_multi_datasets/pretrained.py \
--checkpoint ./transducer_stateless_multi_datasets/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method greedy_search \
--max-sym-per-frame 1 \
/path/to/foo.wav \
/path/to/bar.wav
(2) beam search
./transducer_stateless/pretrained.py \
--checkpoint ./transducer_stateless/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method beam_search \
--beam-size 4 \
/path/to/foo.wav \
/path/to/bar.wav \
./transducer_stateless_multi_datasets/pretrained.py \
--checkpoint ./transducer_stateless_multi_datasets/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method beam_search \
--beam-size 4 \
/path/to/foo.wav \
/path/to/bar.wav
(3) modified beam search
./transducer_stateless/pretrained.py \
--checkpoint ./transducer_stateless/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method modified_beam_search \
--beam-size 4 \
/path/to/foo.wav \
/path/to/bar.wav \
./transducer_stateless_multi_datasets/pretrained.py \
--checkpoint ./transducer_stateless_multi_datasets/exp/pretrained.pt \
--bpe-model ./data/lang_bpe_500/bpe.model \
--method modified_beam_search \
--beam-size 4 \
/path/to/foo.wav \
/path/to/bar.wav
You can also use `./transducer_stateless/exp/epoch-xx.pt`.
You can also use `./transducer_stateless_multi_datasets/exp/epoch-xx.pt`.
Note: ./transducer_stateless/exp/pretrained.pt is generated by
./transducer_stateless/export.py
Note: ./transducer_stateless_multi_datasets/exp/pretrained.pt is generated by
./transducer_stateless_multi_datasets/export.py
"""

2
egs/librispeech/ASR/transducer_stateless_multi_datasets/test_decoder.py
View File

@ -19,7 +19,7 @@
To run this file, do:
cd icefall/egs/librispeech/ASR
python ./transducer_stateless/test_decoder.py
python ./transducer_stateless_multi_datasets/test_decoder.py
"""
import torch