From 3999789c180cd31d74624a7000d14e6d76711873 Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期四, 20 四月 2023 14:51:26 +0800
Subject: [PATCH] docs
---
docs/images/dingding.jpg | 0
egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh | 3
egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md | 26 +++++++-
docs/modescope_pipeline/asr_pipeline.md | 128 ++++++++++++++++++++++++++++++++++++++++--
README.md | 4
5 files changed, 147 insertions(+), 14 deletions(-)
diff --git a/README.md b/README.md
index 03156f3..29ddd4a 100644
--- a/README.md
+++ b/README.md
@@ -12,13 +12,13 @@
[**News**](https://github.com/alibaba-damo-academy/FunASR#whats-new)
| [**Highlights**](#highlights)
| [**Installation**](#installation)
-| [**Docs_EN**](https://alibaba-damo-academy.github.io/FunASR/en/index.html)
+| [**Docs**](https://alibaba-damo-academy.github.io/FunASR/en/index.html)
| [**Tutorial**](https://github.com/alibaba-damo-academy/FunASR/wiki#funasr%E7%94%A8%E6%88%B7%E6%89%8B%E5%86%8C)
| [**Papers**](https://github.com/alibaba-damo-academy/FunASR#citations)
| [**Runtime**](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime)
| [**Model Zoo**](https://github.com/alibaba-damo-academy/FunASR/blob/main/docs/modelscope_models.md)
| [**Contact**](#contact)
-
+|
[**M2MET2.0 Guidence_CN**](https://alibaba-damo-academy.github.io/FunASR/m2met2_cn/index.html)
| [**M2MET2.0 Guidence_EN**](https://alibaba-damo-academy.github.io/FunASR/m2met2/index.html)
diff --git a/docs/images/dingding.jpg b/docs/images/dingding.jpg
index aea2b06..6ac3ab8 100644
--- a/docs/images/dingding.jpg
+++ b/docs/images/dingding.jpg
Binary files differ
diff --git a/docs/modescope_pipeline/asr_pipeline.md b/docs/modescope_pipeline/asr_pipeline.md
index ee4b3ff..645c5d4 100644
--- a/docs/modescope_pipeline/asr_pipeline.md
+++ b/docs/modescope_pipeline/asr_pipeline.md
@@ -1,7 +1,7 @@
# Speech Recognition
> **Note**:
-> The modelscope pipeline supports all the models in [model zoo] to inference and finetine. Here we take model of Paraformer and Paraformer-online as example to demonstrate the usage.
+> The modelscope pipeline supports all the models in [model zoo](https://alibaba-damo-academy.github.io/FunASR/en/modelscope_models.html#pretrained-models-on-modelscope) to inference and finetine. Here we take model of Paraformer and Paraformer-online as example to demonstrate the usage.
## Inference
@@ -33,13 +33,30 @@
# first chunk, 480ms
speech_chunk = speech[0:chunk_stride]
rec_result = inference_pipeline(audio_in=speech_chunk, param_dict=param_dict)
+print(rec_result)
# next chunk, 480ms
speech_chunk = speech[chunk_stride:chunk_stride+chunk_stride]
rec_result = inference_pipeline(audio_in=speech_chunk, param_dict=param_dict)
-
print(rec_result)
```
Full code of demo, please ref to [demo](https://github.com/alibaba-damo-academy/FunASR/discussions/241)
+
+#### [UniASR model](https://www.modelscope.cn/models/damo/speech_UniASR_asr_2pass-zh-cn-8k-common-vocab3445-pytorch-online/summary)
+There are three decoding mode for UniASR model(`fast`銆乣normal`銆乣offline`), for more model detailes, please refer to [docs](https://www.modelscope.cn/models/damo/speech_UniASR_asr_2pass-zh-cn-8k-common-vocab3445-pytorch-online/summary)
+```python
+decoding_model = "fast" # "fast"銆�"normal"銆�"offline"
+inference_pipeline = pipeline(
+ task=Tasks.auto_speech_recognition,
+ model='damo/speech_UniASR_asr_2pass-minnan-16k-common-vocab3825',
+ param_dict={"decoding_model": decoding_model})
+
+rec_result = inference_pipeline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav')
+print(rec_result)
+```
+The decoding mode of `fast` and `normal`
+Full code of demo, please ref to [demo](https://github.com/alibaba-damo-academy/FunASR/discussions/151)
+#### [RNN-T-online model]()
+Undo
#### API-reference
##### define pipeline
@@ -62,19 +79,118 @@
```
In this case of `wav.scp` input, `output_dir` must be set to save the output results
- `audio_fs`: audio sampling rate, only set when audio_in is pcm audio
+- `output_dir`: None (Defalut), the output path of results if set
+### Inference with multi-thread CPUs or multi GPUs
+FunASR also offer recipes [run.sh](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh) to decode with multi-thread CPUs, or multi GPUs.
-### Inference with you data
+- Setting parameters in `infer.sh`
+ - <strong>model:</strong> # model name on ModelScope
+ - <strong>data_dir:</strong> # the dataset dir needs to include `${data_dir}/wav.scp`. If `${data_dir}/text` is also exists, CER will be computed
+ - <strong>output_dir:</strong> # result dir
+ - <strong>batch_size:</strong> # batchsize of inference
+ - <strong>gpu_inference:</strong> # whether to perform gpu decoding, set false for cpu decoding
+ - <strong>gpuid_list:</strong> # set gpus, e.g., gpuid_list="0,1"
+ - <strong>njob:</strong> # the number of jobs for CPU decoding, if `gpu_inference`=false, use CPU decoding, please set `njob`
-### Inference with multi-threads on CPU
+- Decode with multi GPUs:
+```shell
+ bash infer.sh \
+ --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
+ --data_dir "./data/test" \
+ --output_dir "./results" \
+ --batch_size 64 \
+ --gpu_inference true \
+ --gpuid_list "0,1"
+```
+- Decode with multi-thread CPUs:
+```shell
+ bash infer.sh \
+ --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
+ --data_dir "./data/test" \
+ --output_dir "./results" \
+ --gpu_inference false \
+ --njob 64
+```
-### Inference with multi GPU
+- Results
+
+The decoding results can be found in `$output_dir/1best_recog/text.cer`, which includes recognition results of each sample and the CER metric of the whole test set.
+
+If you decode the SpeechIO test sets, you can use textnorm with `stage`=3, and `DETAILS.txt`, `RESULTS.txt` record the results and CER after text normalization.
+
## Finetune with pipeline
### Quick start
+[finetune.py](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/finetune.py)
+```python
+import os
+from modelscope.metainfo import Trainers
+from modelscope.trainers import build_trainer
+from modelscope.msdatasets.audio.asr_dataset import ASRDataset
+
+def modelscope_finetune(params):
+ if not os.path.exists(params.output_dir):
+ os.makedirs(params.output_dir, exist_ok=True)
+ # dataset split ["train", "validation"]
+ ds_dict = ASRDataset.load(params.data_path, namespace='speech_asr')
+ kwargs = dict(
+ model=params.model,
+ data_dir=ds_dict,
+ dataset_type=params.dataset_type,
+ work_dir=params.output_dir,
+ batch_bins=params.batch_bins,
+ max_epoch=params.max_epoch,
+ lr=params.lr)
+ trainer = build_trainer(Trainers.speech_asr_trainer, default_args=kwargs)
+ trainer.train()
+
+
+if __name__ == '__main__':
+ from funasr.utils.modelscope_param import modelscope_args
+ params = modelscope_args(model="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch")
+ params.output_dir = "./checkpoint" # 妯″瀷淇濆瓨璺緞
+ params.data_path = "speech_asr_aishell1_trainsets" # 鏁版嵁璺緞锛屽彲浠ヤ负modelscope涓凡涓婁紶鏁版嵁锛屼篃鍙互鏄湰鍦版暟鎹�
+ params.dataset_type = "small" # 灏忔暟鎹噺璁剧疆small锛岃嫢鏁版嵁閲忓ぇ浜�1000灏忔椂锛岃浣跨敤large
+ params.batch_bins = 2000 # batch size锛屽鏋渄ataset_type="small"锛宐atch_bins鍗曚綅涓篺bank鐗瑰緛甯ф暟锛屽鏋渄ataset_type="large"锛宐atch_bins鍗曚綅涓烘绉掞紝
+ params.max_epoch = 50 # 鏈�澶ц缁冭疆鏁�
+ params.lr = 0.00005 # 璁剧疆瀛︿範鐜�
+
+ modelscope_finetune(params)
+```
+
+```shell
+python finetune.py &> log.txt &
+```
### Finetune with your data
-## Inference with your finetuned model
+- Modify finetune training related parameters in [finetune.py](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/finetune.py)
+ - <strong>output_dir:</strong> # result dir
+ - <strong>data_dir:</strong> # the dataset dir needs to include files: `train/wav.scp`, `train/text`; `validation/wav.scp`, `validation/text`
+ - <strong>dataset_type:</strong> # for dataset larger than 1000 hours, set as `large`, otherwise set as `small`
+ - <strong>batch_bins:</strong> # batch size. For dataset_type is `small`, `batch_bins` indicates the feature frames. For dataset_type is `large`, `batch_bins` indicates the duration in ms
+ - <strong>max_epoch:</strong> # number of training epoch
+ - <strong>lr:</strong> # learning rate
+- Then you can run the pipeline to finetune with:
+```shell
+python finetune.py
+```
+If you want finetune with multi-GPUs, you could:
+```shell
+CUDA_VISIBLE_DEVICES=1,2 python -m torch.distributed.launch --nproc_per_node 2 finetune.py > log.txt 2>&1
+```
+## Inference with your finetuned model
+- Modify inference related parameters in `infer_after_finetune.py`
+ - <strong>modelscope_model_name: </strong> # model name on ModelScope
+ - <strong>output_dir:</strong> # result dir
+ - <strong>data_dir:</strong> # the dataset dir needs to include `test/wav.scp`. If `test/text` is also exists, CER will be computed
+ - <strong>decoding_model_name:</strong> # set the checkpoint name for decoding, e.g., `valid.cer_ctc.ave.pb`
+ - <strong>batch_size:</strong> # batchsize of inference
+
+- Then you can run the pipeline to finetune with:
+```python
+ python infer_after_finetune.py
+```
diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md
index 79cc3c3..c740f71 100644
--- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md
+++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md
@@ -23,21 +23,37 @@
- Setting parameters in `infer.sh`
- <strong>model:</strong> # model name on ModelScope
- - <strong>data_dir:</strong> # the dataset dir needs to include `test/wav.scp`. If `test/text` is also exists, CER will be computed
+ - <strong>data_dir:</strong> # the dataset dir needs to include `${data_dir}/wav.scp`. If `${data_dir}/text` is also exists, CER will be computed
- <strong>output_dir:</strong> # result dir
- <strong>batch_size:</strong> # batchsize of inference
- <strong>gpu_inference:</strong> # whether to perform gpu decoding, set false for cpu decoding
- <strong>gpuid_list:</strong> # set gpus, e.g., gpuid_list="0,1"
- <strong>njob:</strong> # the number of jobs for CPU decoding, if `gpu_inference`=false, use CPU decoding, please set `njob`
-- Then you can run the pipeline to infer with:
-```python
- sh infer.sh
+- Decode with multi GPUs:
+```shell
+ bash infer.sh \
+ --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
+ --data_dir "./data/test" \
+ --output_dir "./results" \
+ --batch_size 64 \
+ --gpu_inference true \
+ --gpuid_list "0,1"
+```
+
+- Decode with multi-thread CPUs:
+```shell
+ bash infer.sh \
+ --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
+ --data_dir "./data/test" \
+ --output_dir "./results" \
+ --gpu_inference false \
+ --njob 64
```
- Results
-The decoding results can be found in `$output_dir/1best_recog/text.cer`, which includes recognition results of each sample and the CER metric of the whole test set.
+The decoding results can be found in `${output_dir}/1best_recog/text.cer`, which includes recognition results of each sample and the CER metric of the whole test set.
If you decode the SpeechIO test sets, you can use textnorm with `stage`=3, and `DETAILS.txt`, `RESULTS.txt` record the results and CER after text normalization.
diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh
index 221479d..b8b011c 100644
--- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh
+++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/infer.sh
@@ -14,8 +14,9 @@
gpuid_list="0,1" # set gpus, e.g., gpuid_list="0,1"
njob=4 # the number of jobs for CPU decoding, if gpu_inference=false, use CPU decoding, please set njob
+. utils/parse_options.sh || exit 1;
-if ${gpu_inference}; then
+if ${gpu_inference} == "true"; then
nj=$(echo $gpuid_list | awk -F "," '{print NF}')
else
nj=$njob
--
Gitblit v1.9.1