diff --git a/docs/source/released_model.md b/docs/source/released_model.md index bd461fd7..baa4ff45 100644 --- a/docs/source/released_model.md +++ b/docs/source/released_model.md @@ -8,8 +8,8 @@ Acoustic Model | Training Data | Token-based | Size | Descriptions | CER | WER | :-------------:| :------------:| :-----: | -----: | :-----: |:-----:| :-----: | :-----: | :-----: [Ds2 Online Aishell ASR0 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr0/asr0_deepspeech2_online_aishell_ckpt_0.2.0.model.tar.gz) | Aishell Dataset | Char-based | 345 MB | 2 Conv + 5 LSTM layers with only forward direction | 0.078 |-| 151 h | [D2 Online Aishell ASR0](../../examples/aishell/asr0) [Ds2 Offline Aishell ASR0 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr0/asr0_deepspeech2_aishell_ckpt_0.1.1.model.tar.gz)| Aishell Dataset | Char-based | 306 MB | 2 Conv + 3 bidirectional GRU layers| 0.064 |-| 151 h | [Ds2 Offline Aishell ASR0](../../examples/aishell/asr0) -[Conformer Online Aishell ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/asr1_chunk_conformer_aishell_ckpt_0.2.0.model.tar.gz) | Aishell Dataset | Char-based | 189 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention| 0.0534 |-| 151 h | [Conformer Online Aishell ASR1](../../examples/aishell/asr1) -[Conformer Offline Aishell ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/asr1_conformer_aishell_ckpt_0.1.2.model.tar.gz) | Aishell Dataset | Char-based | 189 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention rescoring | 0.0483 |-| 151 h | [Conformer Offline Aishell ASR1](../../examples/aishell/asr1) +[Conformer Online Aishell ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/asr1_chunk_conformer_aishell_ckpt_0.2.0.model.tar.gz) | Aishell Dataset | Char-based | 189 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention rescoring| 0.0544 |-| 151 h | [Conformer Online Aishell ASR1](../../examples/aishell/asr1) +[Conformer Offline Aishell ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/asr1_conformer_aishell_ckpt_0.1.2.model.tar.gz) | Aishell Dataset | Char-based | 189 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention rescoring | 0.0464 |-| 151 h | [Conformer Offline Aishell ASR1](../../examples/aishell/asr1) [Transformer Aishell ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/aishell/asr1/asr1_transformer_aishell_ckpt_0.1.1.model.tar.gz) | Aishell Dataset | Char-based | 128 MB | Encoder:Transformer, Decoder:Transformer, Decoding method: Attention rescoring | 0.0523 || 151 h | [Transformer Aishell ASR1](../../examples/aishell/asr1) [Ds2 Offline Librispeech ASR0 Model](https://paddlespeech.bj.bcebos.com/s2t/librispeech/asr0/asr0_deepspeech2_librispeech_ckpt_0.1.1.model.tar.gz)| Librispeech Dataset | Char-based | 518 MB | 2 Conv + 3 bidirectional LSTM layers| - |0.0725| 960 h | [Ds2 Offline Librispeech ASR0](../../examples/librispeech/asr0) [Conformer Librispeech ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/librispeech/asr1/asr1_conformer_librispeech_ckpt_0.1.1.model.tar.gz) | Librispeech Dataset | subword-based | 191 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention rescoring |-| 0.0337 | 960 h | [Conformer Librispeech ASR1](../../examples/librispeech/asr1) diff --git a/examples/aishell/asr1/RESULTS.md b/examples/aishell/asr1/RESULTS.md index 7730baf1..d8125e62 100644 --- a/examples/aishell/asr1/RESULTS.md +++ b/examples/aishell/asr1/RESULTS.md @@ -18,7 +18,7 @@ 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 | test | attention | 16, -1 | - | 0.0534 | +| conformer | 47.06M | conf/chunk_conformer.yaml | spec_aug | test | attention | 16, -1 | - | 0.0551 | | conformer | 47.06M | conf/chunk_conformer.yaml | spec_aug | test | ctc_greedy_search | 16, -1 | - | 0.0629 | | conformer | 47.06M | conf/chunk_conformer.yaml | spec_aug | test | ctc_prefix_beam_search | 16, -1 | - | 0.0629 | | conformer | 47.06M | conf/chunk_conformer.yaml | spec_aug | test | attention_rescoring | 16, -1 | - | 0.0544 | diff --git a/examples/aishell/asr1/local/test.sh b/examples/aishell/asr1/local/test.sh index a88feeed..26926b4a 100755 --- a/examples/aishell/asr1/local/test.sh +++ b/examples/aishell/asr1/local/test.sh @@ -84,7 +84,7 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then exit 1 fi python utils/format_rsl.py \ - --origin_hyp ${output_dir}/${type}.rsl + --origin_hyp ${output_dir}/${type}.rsl \ --trans_hyp ${output_dir}/${type}.rsl.text python utils/compute-wer.py --char=1 --v=1 \ data/manifest.test.text ${output_dir}/${type}.rsl.text > ${output_dir}/${type}.error @@ -100,7 +100,7 @@ if [ ${stage} -le 101 ] && [ ${stop_stage} -ge 101 ]; then output_dir=${ckpt_prefix} for type in attention ctc_greedy_search ctc_prefix_beam_search attention_rescoring; do python utils/format_rsl.py \ - --origin_hyp ${output_dir}/${type}.rsl + --origin_hyp ${output_dir}/${type}.rsl \ --trans_hyp_sclite ${output_dir}/${type}.rsl.text.sclite mkdir -p ${output_dir}/${type}_sclite diff --git a/examples/voxceleb/sv0/README.md b/examples/voxceleb/sv0/README.md new file mode 100644 index 00000000..567963e5 --- /dev/null +++ b/examples/voxceleb/sv0/README.md @@ -0,0 +1,151 @@ +# ECAPA-TDNN with VoxCeleb +This example contains code used to train a ECAPA-TDNN model with [VoxCeleb dataset](https://www.robots.ox.ac.uk/~vgg/data/voxceleb/index.html#about) + +## Overview +All the scripts you need are in the `run.sh`. There are several stages in the `run.sh`, and each stage has its function. +| Stage | Function | +|:---- |:----------------------------------------------------------- | +| 0 | Process data. It includes:
(1) Download the VoxCeleb1 dataset
(2) Download the VoxCeleb2 dataset
(3) Convert the VoxCeleb2 m4a to wav format
(4) Get the manifest files of the train, development and test dataset
(5) Download the RIR Noise dataset and Get the noise manifest files for augmentation | +| 1 | Train the model | +| 2 | Test the speaker verification with VoxCeleb trial| + +You can choose to run a range of stages by setting the `stage` and `stop_stage `. + +For example, if you want to execute the code in stage 1 and stage 2, you can run this script: +```bash +bash run.sh --stage 1 --stop_stage 2 +``` +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 +bash run.sh --stage 1 --stop_stage 1 +``` +The document below will describe the scripts in the `run.sh` in detail. +## The environment variables +The path.sh contains the environment variable. +```bash +source path.sh +``` +This script needs to be run first. + +And another script is also needed: +```bash +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 the `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. +`exp_dir` denotes the experiment directory, e.g. "exp/ecapa-tdnn-vox12-big/" + +You can set the local variables when you use the `run.sh` + +For example, you can set the `gpus` when you use the command line.: +```bash +bash run.sh --gpus 0,1 +``` +## Stage 0: Data processing +To use this example, you need to process data firstly and you can use stage 0 in the `run.sh` to do this. The code is shown below: + +```bash + 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 +bash run.sh --stage 0 --stop_stage 0 +``` +You can also just run these scripts in your command line. +```bash +source path.sh +bash ./local/data.sh +``` +After processing the data, the `data` directory will look like this: +```bash +data/ +├── rir_noise +│   ├── csv +│   │   ├── noise.csv +│   │   └── rir.csv +│   ├── manifest.pointsource_noises +│   ├── manifest.real_rirs_isotropic_noises +│   └── manifest.simulated_rirs +├── vox +│   ├── csv +│   │   ├── dev.csv +│   │   ├── enroll.csv +│   │   ├── test.csv +│   │   └── train.csv +│   └── meta +│   └── label2id.txt +└── vox1 + ├── list_test_all2.txt + ├── list_test_all.txt + ├── list_test_hard2.txt + ├── list_test_hard.txt + ├── manifest.dev + ├── manifest.test + ├── veri_test2.txt + ├── veri_test.txt + ├── voxceleb1.dev.meta + └── voxceleb1.test.meta +``` +## Stage 1: Model training +If you want to train the model. you can use stage 1 in the `run.sh`. The code is shown below. +```bash +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 +bash run.sh --stage 0 --stop_stage 1 +``` +or you can run these scripts in the command line (only use CPU). +```bash +source path.sh +bash ./local/data.sh ./data/ conf/ecapa_tdnn.yaml +CUDA_VISIBLE_DEVICES= ./local/train.sh ./data/ exp/ecapa-tdnn-vox12-big/ conf/ecapa_tdnn.yaml +``` +## Stage 2: Model Testing +The test stage is to evaluate the model performance. The code of the test stage is shown below: +```bash + if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then + # test ckpt avg_n + CUDA_VISIBLE_DEVICES=0 ./local/test.sh ${dir} ${exp_dir} ${conf_path} || 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 and stage 2: +```bash +bash run.sh --stage 0 --stop_stage 2 +``` +or you can run these scripts in the command line (only use CPU). +```bash +source path.sh +bash ./local/data.sh ./data/ conf/ecapa_tdnn.yaml +CUDA_VISIBLE_DEVICES= ./local/train.sh ./data/ exp/ecapa-tdnn-vox12-big/ conf/ecapa_tdnn.yaml +CUDA_VISIBLE_DEVICES= ./local/test.sh ./data/ exp/ecapa-tdnn-vox12-big/ conf/ecapa_tdnn.yaml +``` + +## 3: Pretrained Model +You can get the pretrained models from [this](../../../docs/source/released_model.md). + +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/vector/voxceleb/sv0_ecapa_tdnn_voxceleb12_ckpt_0_2_0.tar.gz +tar xzvf sv0_ecapa_tdnn_voxceleb12_ckpt_0_2_0.tar.gz +source path.sh +# If you have processed the data and get the manifest file， you can skip the following 2 steps + +CUDA_VISIBLE_DEVICES= ./local/test.sh ./data sv0_ecapa_tdnn_voxceleb12_ckpt_0_1_2 conf/ecapa_tdnn.yaml +``` +The performance of the released models are shown in [this](./RESULTS.md)