From fa724285f3b799b97b4348ad3b1084afc0764f9b Mon Sep 17 00:00:00 2001
From: zxcd <228587199@qq.com>
Date: Fri, 6 Jan 2023 07:25:57 +0000
Subject: [PATCH] add readme and result, and fix some bug.
---
dataset/tal_cs/README.md | 12 +-
dataset/tal_cs/tal_cs.py | 54 ++++----
docs/source/released_model.md | 3 +-
examples/tal_cs/asr1/README.md | 190 +++++++++++++++++++++++++++++
examples/tal_cs/asr1/RESULTS.md | 11 ++
examples/tal_cs/asr1/local/data.sh | 4 +-
examples/tal_cs/asr1/run.sh | 10 +-
7 files changed, 250 insertions(+), 34 deletions(-)
create mode 100644 examples/tal_cs/asr1/README.md
create mode 100644 examples/tal_cs/asr1/RESULTS.md
diff --git a/dataset/tal_cs/README.md b/dataset/tal_cs/README.md
index a04f7411f..633056360 100644
--- a/dataset/tal_cs/README.md
+++ b/dataset/tal_cs/README.md
@@ -1,3 +1,13 @@
# [TAL_CSASR](https://ai.100tal.com/dataset/)
-This data set is TAL English class audio, including mixed Chinese and English speech. Each audio has only one speaker, and this data set has more than 100 speakers. (File 63.36G) This data contains the sample of intra sentence and inter sentence mixing as shown in Figure 1. The ratio between Chinese characters and English words in the data is 13:1.
+This data set is TAL English class audio, including mixed Chinese and English speech. Each audio has only one speaker, and this data set has more than 100 speakers. (File 63.36G) This data contains the sample of intra sentence and inter sentence mixing. The ratio between Chinese characters and English words in the data is 13:1.
+
+- Total data: 587H (train_set: 555.9H, dev_set: 8H, test_set: 23.6H)
+- Sample rate: 16000
+- Sample bit: 16
+- Recording device: microphone
+- Speaker number: 200+
+- Recording time: 2019
+- Data format: audio: .wav; test: .txt
+- Audio duration: 1-60s
+- Data type: audio of English teachers' teaching
diff --git a/dataset/tal_cs/tal_cs.py b/dataset/tal_cs/tal_cs.py
index 8302e45de..2024b21e3 100644
--- a/dataset/tal_cs/tal_cs.py
+++ b/dataset/tal_cs/tal_cs.py
@@ -1,4 +1,4 @@
-# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
@@ -59,33 +59,31 @@ def create_manifest(data_dir, manifest_path):
wav_dir = os.path.join(data_dir, 'wav')
text_filepath = os.path.join(data_dir, 'label.txt')
for subfolder, _, filelist in sorted(os.walk(wav_dir)):
- text_filelist = text_filepath
- if len(text_filelist) > 0:
- for line in io.open(text_filepath, encoding="utf8"):
- segments = line.strip().split()
- nchars = len(segments[1:])
- text = ' '.join(segments[1:]).lower()
-
- audio_filepath = os.path.abspath(
- os.path.join(subfolder, segments[0] + '.wav'))
- audio_data, samplerate = soundfile.read(audio_filepath)
- duration = float(len(audio_data)) / samplerate
-
- utt = os.path.splitext(os.path.basename(audio_filepath))[0]
- utt2spk = '-'.join(utt.split('-')[:2])
-
- json_lines.append(
- json.dumps({
- 'utt': utt,
- 'utt2spk': utt2spk,
- 'feat': audio_filepath,
- 'feat_shape': (duration, ), # second
- 'text': text,
- }))
-
- total_sec += duration
- total_char += nchars
- total_num += 1
+ for line in io.open(text_filepath, encoding="utf8"):
+ segments = line.strip().split()
+ nchars = len(segments[1:])
+ text = ' '.join(segments[1:]).lower()
+
+ audio_filepath = os.path.abspath(
+ os.path.join(subfolder, segments[0] + '.wav'))
+ audio_data, samplerate = soundfile.read(audio_filepath)
+ duration = float(len(audio_data)) / samplerate
+
+ utt = os.path.splitext(os.path.basename(audio_filepath))[0]
+ utt2spk = '-'.join(utt.split('-')[:2])
+
+ json_lines.append(
+ json.dumps({
+ 'utt': utt,
+ 'utt2spk': utt2spk,
+ 'feat': audio_filepath,
+ 'feat_shape': (duration, ), # second
+ 'text': text,
+ }))
+
+ total_sec += duration
+ total_char += nchars
+ total_num += 1
with codecs.open(manifest_path, 'w', 'utf-8') as out_file:
for line in json_lines:
diff --git a/docs/source/released_model.md b/docs/source/released_model.md
index 87c58b787..10a39e239 100644
--- a/docs/source/released_model.md
+++ b/docs/source/released_model.md
@@ -17,6 +17,7 @@ Acoustic Model | Training Data | Token-based | Size | Descriptions | CER | WER |
[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 |-| 0.0338 | 960 h | [Conformer Librispeech ASR1](../../examples/librispeech/asr1) | python |
[Transformer Librispeech ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/librispeech/asr1/asr1_transformer_librispeech_ckpt_0.1.1.model.tar.gz) | Librispeech Dataset | subword-based | 131 MB | Encoder:Transformer, Decoder:Transformer, Decoding method: Attention rescoring |-| 0.0381 | 960 h | [Transformer Librispeech ASR1](../../examples/librispeech/asr1) | python |
[Transformer Librispeech ASR2 Model](https://paddlespeech.bj.bcebos.com/s2t/librispeech/asr2/asr2_transformer_librispeech_ckpt_0.1.1.model.tar.gz) | Librispeech Dataset | subword-based | 131 MB | Encoder:Transformer, Decoder:Transformer, Decoding method: JoinCTC w/ LM |-| 0.0240 | 960 h | [Transformer Librispeech ASR2](../../examples/librispeech/asr2) | python |
+[Conformer TALCS ASR1 Model](https://paddlespeech.bj.bcebos.com/s2t/tal_cs/asr1/asr1_conformer_talcs_ckpt_1.4.0.model.tar.gz) | TALCS Dataset | subword-based | 470 MB | Encoder:Conformer, Decoder:Transformer, Decoding method: Attention rescoring |-| 0.0844 | 587 h | [Conformer TALCS ASR1](../../examples/tal_cs/asr1) | python |
### Self-Supervised Pre-trained Model
Model | Pre-Train Method | Pre-Train Data | Finetune Data | Size | Descriptions | CER | WER | Example Link |
@@ -29,7 +30,7 @@ Model | Pre-Train Method | Pre-Train Data | Finetune Data | Size | Descriptions
### Whisper Model
Demo Link | Training Data | Size | Descriptions | CER | Model
:-----------: | :-----:| :-------: | :-----: | :-----: |:---------:|
-[Whisper](../../demos/whisper) | 680kh from internet | large: 5.8G,</br>medium: 2.9G,</br>small: 923M,</br>base: 277M,</br>tiny: 145M | Encoder:Transformer,</br> Decoder:Transformer, </br>Decoding method: </br>Greedy search | 2.7 </br>(large, Librispeech) | [whisper-large](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-large-model.tar.gz) </br>[whisper-medium](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-medium-model.tar.gz) </br>[whisper-medium-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-medium-en-model.tar.gz) </br>[whisper-small](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-small-model.tar.gz) </br>[whisper-small-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-small-en-model.tar.gz) </br>[whisper-base](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-base-model.tar.gz) </br>[whisper-base-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-base-en-model.tar.gz) </br>[whisper-tiny](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-tiny-model.tar.gz) </br>[whisper-tiny-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-tiny-en-model.tar.gz)
+[Whisper](../../demos/whisper) | 680kh from internet | large: 5.8G,</br>medium: 2.9G,</br>small: 923M,</br>base: 277M,</br>tiny: 145M | Encoder:Transformer,</br> Decoder:Transformer, </br>Decoding method: </br>Greedy search | 0.027 </br>(large, Librispeech) | [whisper-large](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-large-model.tar.gz) </br>[whisper-medium](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-medium-model.tar.gz) </br>[whisper-medium-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-medium-en-model.tar.gz) </br>[whisper-small](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-small-model.tar.gz) </br>[whisper-small-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-small-en-model.tar.gz) </br>[whisper-base](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-base-model.tar.gz) </br>[whisper-base-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-base-en-model.tar.gz) </br>[whisper-tiny](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-tiny-model.tar.gz) </br>[whisper-tiny-English-only](https://paddlespeech.bj.bcebos.com/whisper/whisper_model_20221122/whisper-tiny-en-model.tar.gz)
### Language Model based on NGram
|Language Model | Training Data | Token-based | Size | Descriptions|
diff --git a/examples/tal_cs/asr1/README.md b/examples/tal_cs/asr1/README.md
new file mode 100644
index 000000000..83a27ac1e
--- /dev/null
+++ b/examples/tal_cs/asr1/README.md
@@ -0,0 +1,190 @@
+# Transformer/Conformer ASR with TALCS
+This example contains code used to train [u2](https://arxiv.org/pdf/2012.05481.pdf) model (Transformer or [Conformer](https://arxiv.org/pdf/2005.08100.pdf) model) with [TALCS dataset](https://ai.100tal.com/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: <br> (1) Download the dataset <br> (2) Calculate the CMVN of the train dataset <br> (3) Get the vocabulary file <br> (4) Get the manifest files of the train, development and test dataset<br> (5) Get the sentencepiece model |
+| 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
+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
+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.
+```bash
+. ./path.sh
+. ./cmd.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 `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 stages 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
+bash run.sh --gpus 0,1 --avg_num 10
+```
+## 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:
+```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
+. ./path.sh
+. ./cmd.sh
+bash ./local/data.sh
+```
+After processing the data, the `data` directory will look like this:
+```bash
+data/
+|-- dev_set.meta
+|-- lang_char
+| `-- bpe_bpe_11297.model
+| `-- bpe_bpe_11297.vocab
+| `-- vocab.txt
+|-- manifest.dev
+|-- manifest.dev.raw
+|-- manifest.test
+|-- manifest.test.raw
+|-- manifest.train
+|-- manifest.train.raw
+|-- mean_std.json
+|-- test_set.meta
+`-- train_set.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.
+```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
+. ./path.sh
+. ./cmd.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:
+```bash
+ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
+ # avg n best model
+ avg.sh best exp/${ckpt}/checkpoints ${avg_num}
+ fi
+```
+The `avg.sh` is 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
+bash run.sh --stage 0 --stop_stage 2
+```
+or you can run these scripts in the command line (only use CPU).
+
+```bash
+. ./path.sh
+. ./cmd.sh
+bash ./local/data.sh
+CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer
+avg.sh best exp/conformer/checkpoints 10
+```
+## Stage 3: Model Testing
+The test stage is to evaluate the model performance. The code of test stage is shown below:
+```bash
+ 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
+bash run.sh --stage 0 --stop_stage 3
+```
+or you can run these scripts in the command line (only use CPU).
+```bash
+. ./path.sh
+. ./cmd.sh
+bash ./local/data.sh
+CUDA_VISIBLE_DEVICES= ./local/train.sh conf/conformer.yaml conformer
+avg.sh best exp/conformer/checkpoints 10
+CUDA_VISIBLE_DEVICES= ./local/test.sh conf/conformer.yaml exp/conformer/checkpoints/avg_10
+```
+## Pretrained Model
+You can get the pretrained transformer or conformer 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:
+```bash
+wget https://paddlespeech.bj.bcebos.com/s2t/tal_cs/asr1/asr1_conformer_talcs_ckpt_1.4.0.model.tar.gz
+tar xzvf asr1_conformer_talcs_ckpt_1.4.0.model.tar.gz
+source path.sh
+# If you have process the data and get the manifest fileļ¼ 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/conformer.yaml exp/conformer/checkpoints/avg_10
+```
+The performance of the released models are shown in [here](./RESULTS.md).
+
+## 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
+```bash
+ 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} ${audio_file} || exit -1
+ fi
+```
+you can train the model by yourself using ```bash run.sh --stage 0 --stop_stage 3```, or you can download the pretrained model through the script below:
+```bash
+wget https://paddlespeech.bj.bcebos.com/s2t/tal_cs/asr1/asr1_conformer_talcs_ckpt_1.4.0.model.tar.gz
+tar xzvf asr1_conformer_talcs_ckpt_1.4.0.model.tar.gz
+```
+You can download the audio demo:
+```bash
+wget -nc https://paddlespeech.bj.bcebos.com/datasets/single_wav/zh/demo_01_03.wav -P data/
+```
+You need to prepare an audio file or use the audio demo above, please confirm the sample rate of the audio is 16K. You can get the result of the audio demo by running the script below.
+```bash
+CUDA_VISIBLE_DEVICES= ./local/test_wav.sh conf/conformer.yaml exp/conformer/checkpoints/avg_10 data/demo_01_03.wav
+```
diff --git a/examples/tal_cs/asr1/RESULTS.md b/examples/tal_cs/asr1/RESULTS.md
new file mode 100644
index 000000000..e207ab783
--- /dev/null
+++ b/examples/tal_cs/asr1/RESULTS.md
@@ -0,0 +1,11 @@
+# TALCS
+
+## Conformer
+train: Epoch 100, 3 V100-32G, best avg: 10
+
+| Model | Params | Config | Augmentation| Test set | Decode method | Loss | WER |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| conformer | 47.63 M | conf/conformer.yaml | spec_aug | test-set | attention | 9.85091028213501 | 0.102786 |
+| conformer | 47.63 M | conf/conformer.yaml | spec_aug | test-set | ctc_greedy_search | 9.85091028213501 | 0.103538 |
+| conformer | 47.63 M | conf/conformer.yaml | spec_aug | test-set | ctc_prefix_beam_search | 9.85091028213501 | 0.103317 |
+| conformer | 47.63 M | conf/conformer.yaml | spec_aug | test-set | attention_rescoring | 9.85091028213501 | 0.084374 |
diff --git a/examples/tal_cs/asr1/local/data.sh b/examples/tal_cs/asr1/local/data.sh
index 6b8040865..7ea12809f 100644
--- a/examples/tal_cs/asr1/local/data.sh
+++ b/examples/tal_cs/asr1/local/data.sh
@@ -8,8 +8,8 @@ nbpe=11297
bpemode=bpe
bpeprefix="${dict_dir}/bpe_${bpemode}_${nbpe}"
-stride_ms=10
-window_ms=25
+stride_ms=20
+window_ms=30
sample_rate=16000
feat_dim=80
diff --git a/examples/tal_cs/asr1/run.sh b/examples/tal_cs/asr1/run.sh
index 3b8ffa19f..32fd69693 100644
--- a/examples/tal_cs/asr1/run.sh
+++ b/examples/tal_cs/asr1/run.sh
@@ -2,9 +2,9 @@
source path.sh || exit 1;
set -e
-gpus=1,2,3
+gpus=0,1,2,3
stage=0
-stop_stage=100
+stop_stage=4
conf_path=conf/conformer.yaml
ips= #xxx.xxx.xxx.xxx,xxx.xxx.xxx.xxx
decode_conf_path=conf/tuning/decode.yaml
@@ -43,3 +43,9 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# test a single .wav file
CUDA_VISIBLE_DEVICES=0 ./local/test_wav.sh ${conf_path} ${decode_conf_path} exp/${ckpt}/checkpoints/${avg_ckpt} ${audio_file} || exit -1
fi
+
+# Not supported at now!!!
+if [ ${stage} -le 51 ] && [ ${stop_stage} -ge 51 ]; then
+ # export ckpt avg_n
+ CUDA_VISIBLE_DEVICES=0 ./local/export.sh ${conf_path} exp/${ckpt}/checkpoints/${avg_ckpt} exp/${ckpt}/checkpoints/${avg_ckpt}.jit
+fi
\ No newline at end of file