PaddleSpeech/examples/aishell/asr1/README.md

Transformer/Conformer ASR with Aishell

This example contains code used to train a Transformer or Conformer model with Aishell dataset

Overview

All the scripts you need are in run.sh. There are several stages in run.sh, and each stage has its function.

Stage	Function
0	Process data. It includes: (1) Download the dataset (2) Calculate the CMVN of the train dataset (3) Get the vocabulary file (4) Get the manifest files of the train, development and test dataset
1	Train the model
2	Get the final model by averaging the top-k models, set k = 1 means to choose the best model
3	Test the final model performance
4	Get ctc alignment of test data using the final model
5	Infer the single audio file

You can choose to run a range of stages by setting stage and stop_stage .

For example, if you want to execute the code in stage 2 and stage 3, you can run this script:

bash run.sh --stage 2 --stop_stage 3

Or you can set stage equal to stop-stage to only run one stage. For example, if you only want to run stage 0, you can use the script below:

bash run.sh --stage 0 --stop_stage 0

The document below will describe the scripts in run.sh in detail.

The Environment Variables

The path.sh contains the environment variables.

source path.sh

This script needs to be run first. And another script is also needed:

source ${MAIN_ROOT}/utils/parse_options.sh

It will support the way of using --variable value in the shell scripts.

The Local Variables

Some local variables are set in run.sh. gpus denotes the GPU number you want to use. If you set gpus=, it means you only use CPU. stage denotes the number of the stage you want to start from in the experiments. stop stage denotes the number of the stage you want to end at in the experiments. conf_path denotes the config path of the model. avg_num denotes the number K of top-K models you want to average to get the final model. audio_file denotes the file path of the single file you want to infer in stage 5 ckpt denotes the checkpoint prefix of the model, e.g. "conformer" You can set the local variables (except ckpt) when you use run.sh

For example, you can set the gpus and avg_num when you use the command line.:

bash run.sh --gpus 0,1 --avg_num 20

Stage 0: Data Processing

To use this example, you need to process data firstly and you can use stage 0 in run.sh to do this. The code is shown below:

 if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
     # prepare data
     bash ./local/data.sh || exit -1
 fi

Stage 0 is for processing the data.

If you only want to process the data. You can run

bash run.sh --stage 0 --stop_stage 0

You can also just run these scripts in your command line.

source path.sh
bash ./local/data.sh

After processing the data, the data directory will look like this:

data/
|-- dev.meta
|-- lang_char
|   `-- vocab.txt
|-- manifest.dev
|-- manifest.dev.raw
|-- manifest.test
|-- manifest.test.raw
|-- manifest.train
|-- manifest.train.raw
|-- mean_std.json
|-- test.meta
`-- train.meta

Stage 1: Model Training

If you want to train the model. you can use stage 1 in run.sh. The code is shown below.

if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
     # train model, all `ckpt` under `exp` dir
     CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${conf_path} ${ckpt}
 fi

If you want to train the model, you can use the script below to execute stage 0 and stage 1:

bash run.sh --stage 0 --stop_stage 1

or you can run these scripts in the command line (only use CPU).

source path.sh
bash ./local/data.sh
CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer

Stage 2: Top-k Models Averaging

After training the model, we need to get the final model for testing and inference. In every epoch, the model checkpoint is saved, so we can choose the best model from them based on the validation loss or we can sort them and average the parameters of the top-k models to get the final model. We can use stage 2 to do this, and the code is shown below:

 if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
     # avg n best model
     avg.sh best exp/${ckpt}/checkpoints ${avg_num}
 fi

The avg.shis in the ../../../utils/ which is define in the path.sh. If you want to get the final model, you can use the script below to execute stage 0, stage 1, and stage 2:

bash run.sh --stage 0 --stop_stage 2

or you can run these scripts in the command line (only use CPU).

source path.sh
bash ./local/data.sh
CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer
avg.sh best exp/conformer/checkpoints 20

Stage 3: Model Testing

The test stage is to evaluate the model performance. The code of the test stage is shown below:

 if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
     # test ckpt avg_n
     CUDA_VISIBLE_DEVICES=0 ./local/test.sh ${conf_path} exp/${ckpt}/checkpoints/${avg_ckpt} || exit -1
 fi

If you want to train a model and test it, you can use the script below to execute stage 0, stage 1, stage 2, and stage 3 :

bash run.sh --stage 0 --stop_stage 3

or you can run these scripts in the command line (only use CPU).

source path.sh
bash ./local/data.sh
CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer
avg.sh best exp/conformer/checkpoints 20
CUDA_VISIBLE_DEVICES= ./local/test.sh conf/conformer.yaml exp/conformer/checkpoints/avg_20

Pretrained Model

You can get the pretrained transformer or conformer using the scripts below:

# Conformer:
wget https://deepspeech.bj.bcebos.com/release2.1/aishell/s1/aishell.release.tar.gz

# Chunk Conformer:
wget https://deepspeech.bj.bcebos.com/release2.1/aishell/s1/aishell.chunk.release.tar.gz

# Transformer:
wget https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/transformer.model.tar.gz

using the tar scripts to unpack the model and then you can use the script to test the model.

For example:

wget https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/transformer.model.tar.gz
tar xzvf transformer.model.tar.gz
source path.sh
# If you have process the data and get the manifest file， you can skip the following 2 steps
bash local/data.sh --stage -1 --stop_stage -1
bash local/data.sh --stage 2 --stop_stage 2
CUDA_VISIBLE_DEVICES= ./local/test.sh conf/transformer.yaml exp/transformer/checkpoints/avg_20

The performance of the released models are shown below:

Conformer

Model	Params	Config	Augmentation	Test set	Decode method	Loss	CER
conformer	47.07M	conf/conformer.yaml	spec_aug + shift	test	attention	-	0.059858
conformer	47.07M	conf/conformer.yaml	spec_aug + shift	test	ctc_greedy_search	-	0.062311
conformer	47.07M	conf/conformer.yaml	spec_aug + shift	test	ctc_prefix_beam_search	-	0.062196
conformer	47.07M	conf/conformer.yaml	spec_aug + shift	test	attention_rescoring	-	0.054694

Chunk Conformer

Need set decoding.decoding_chunk_size=16 when decoding.

Model	Params	Config	Augmentation	Test set	Decode method	Chunk Size & Left Chunks	Loss	CER
conformer	47.06M	conf/chunk_conformer.yaml	spec_aug + shift	test	attention	16, -1	-	0.061939
conformer	47.06M	conf/chunk_conformer.yaml	spec_aug + shift	test	ctc_greedy_search	16, -1	-	0.070806
conformer	47.06M	conf/chunk_conformer.yaml	spec_aug + shift	test	ctc_prefix_beam_search	16, -1	-	0.070739
conformer	47.06M	conf/chunk_conformer.yaml	spec_aug + shift	test	attention_rescoring	16, -1	-	0.059400

Transformer

Model	Params	Config	Augmentation	Test set	Decode method	Loss	CER
transformer	31.95M	conf/transformer.yaml	spec_aug	test	attention	3.858648955821991	0.057293
transformer	31.95M	conf/transformer.yaml	spec_aug	test	ctc_greedy_search	3.858648955821991	0.061837
transformer	31.95M	conf/transformer.yaml	spec_aug	test	ctc_prefix_beam_search	3.858648955821991	0.061685
transformer	31.95M	conf/transformer.yaml	spec_aug	test	attention_rescoring	3.858648955821991	0.053844

Stage 4: CTC Alignment

If you want to get the alignment between the audio and the text, you can use the ctc alignment. The code of this stage is shown below:

 if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
     # ctc alignment of test data
     CUDA_VISIBLE_DEVICES=0 ./local/align.sh ${conf_path} exp/${ckpt}/checkpoints/${avg_ckpt} || exit -1
 fi

If you want to train the model, test it and do the alignment, you can use the script below to execute stage 0, stage 1, stage 2, and stage 3 :

bash run.sh --stage 0 --stop_stage 4

or if you only need to train a model and do the alignment, you can use these scripts to escape stage 3(test stage):

bash run.sh --stage 0 --stop_stage 2
bash run.sh --stage 4 --stop_stage 4

or you can also use these scripts in the command line (only use CPU).

source path.sh
bash ./local/data.sh
CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer
avg.sh best exp/conformer/checkpoints 20
# test stage is optional
CUDA_VISIBLE_DEVICES= ./local/test.sh conf/conformer.yaml exp/conformer/checkpoints/avg_20
CUDA_VISIBLE_DEVICES= ./local/align.sh conf/conformer.yaml exp/conformer/checkpoints/avg_20

Stage 5: Single Audio File Inference

In some situations, you want to use the trained model to do the inference for the single audio file. You can use stage 5. The code is shown below

 if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
     # test a single .wav file
     CUDA_VISIBLE_DEVICES=0 ./local/test_wav.sh ${conf_path} exp/${ckpt}/checkpoints/${avg_ckpt} ${audio_file} || exit -1
 fi

you can train the model by yourself using bash run.sh --stage 0 --stop_stage 3, or you can download the pretrained model through the script below:

wget https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/transformer.model.tar.gz
tar xzvf transformer.model.tar.gz

You can download the audio demo:

wget -nc https://paddlespeech.bj.bcebos.com/datasets/single_wav/zh/demo_01_03.wav -P data/

You need to prepare an audio file or use the audio demo above, please confirm the sample rate of the audio is 16K. You can get the result by running the script below.

CUDA_VISIBLE_DEVICES= ./local/test_wav.sh conf/transformer.yaml exp/transformer/checkpoints/avg_20 data/demo_01_03.wav