From 33d3d2084403fd34b79c835d2f2fe04f6cd8f738 Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期三, 13 九月 2023 09:33:54 +0800
Subject: [PATCH] Merge branch 'main' of github.com:alibaba-damo-academy/FunASR add

---
 egs_modelscope/asr/TEMPLATE/README.md |  223 +++++++++++++++++++++++++++++++++++++++----------------
 1 files changed, 158 insertions(+), 65 deletions(-)

diff --git a/egs_modelscope/asr/TEMPLATE/README.md b/egs_modelscope/asr/TEMPLATE/README.md
index 8b6b24d..a8cb486 100644
--- a/egs_modelscope/asr/TEMPLATE/README.md
+++ b/egs_modelscope/asr/TEMPLATE/README.md
@@ -1,12 +1,14 @@
+([绠�浣撲腑鏂嘳(./README_zh.md)|English)
+
 # Speech Recognition
 
-> **Note**: 
-> 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.
+> **Note**:
+> The modelscope pipeline supports all the models in [model zoo](https://alibaba-damo-academy.github.io/FunASR/en/model_zoo/modelscope_models.html#pretrained-models-on-modelscope) to inference and finetine. Here we take the typic models as examples to demonstrate the usage.
 
 ## Inference
 
 ### Quick start
-#### [Paraformer model](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/summary)
+#### [Paraformer Model](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/summary)
 ```python
 from modelscope.pipelines import pipeline
 from modelscope.utils.constant import Tasks
@@ -19,30 +21,52 @@
 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)
 ```
-#### [Paraformer-online model](https://www.modelscope.cn/models/damo/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online/summary)
+#### [Paraformer-online Model](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/summary)
+##### Streaming Decoding
 ```python
 inference_pipeline = pipeline(
     task=Tasks.auto_speech_recognition,
-    model='damo/speech_paraformer_asr_nat-zh-cn-16k-common-vocab8404-online',
+    model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online',
+    model_revision='v1.0.6',
+    update_model=False,
+    mode='paraformer_streaming'
     )
 import soundfile
 speech, sample_rate = soundfile.read("example/asr_example.wav")
 
-param_dict = {"cache": dict(), "is_final": False}
-chunk_stride = 7680# 480ms
-# first chunk, 480ms
-speech_chunk = speech[0:chunk_stride] 
+chunk_size = [5, 10, 5] #[5, 10, 5] 600ms, [8, 8, 4] 480ms
+param_dict = {"cache": dict(), "is_final": False, "chunk_size": chunk_size}
+chunk_stride = chunk_size[1] * 960 # 600ms銆�480ms
+# first chunk, 600ms
+speech_chunk = speech[0:chunk_stride]
 rec_result = inference_pipeline(audio_in=speech_chunk, param_dict=param_dict)
 print(rec_result)
-# next chunk, 480ms
+# next chunk, 600ms
 speech_chunk = speech[chunk_stride:chunk_stride+chunk_stride]
 rec_result = inference_pipeline(audio_in=speech_chunk, param_dict=param_dict)
 print(rec_result)
 ```
+
+##### Fake Streaming Decoding
+```python
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipeline = pipeline(
+    task=Tasks.auto_speech_recognition,
+    model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online',
+    model_revision='v1.0.6',
+    update_model=False,
+    mode="paraformer_fake_streaming"
+)
+audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav'
+rec_result = inference_pipeline(audio_in=audio_in)
+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)
+#### [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 details, 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(
@@ -53,47 +77,67 @@
 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`
+The decoding mode of `fast` and `normal` is fake streaming, which could be used for evaluating of recognition accuracy.
 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
+#### [MFCCA Model](https://www.modelscope.cn/models/NPU-ASLP/speech_mfcca_asr-zh-cn-16k-alimeeting-vocab4950/summary)
+For more model details, please refer to [docs](https://www.modelscope.cn/models/NPU-ASLP/speech_mfcca_asr-zh-cn-16k-alimeeting-vocab4950/summary)
+```python
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipeline = pipeline(
+    task=Tasks.auto_speech_recognition,
+    model='NPU-ASLP/speech_mfcca_asr-zh-cn-16k-alimeeting-vocab4950',
+    model_revision='v3.0.0'
+)
+
+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)
+```
+
+### API-reference
+#### Define pipeline
 - `task`: `Tasks.auto_speech_recognition`
-- `model`: model name in [model zoo](https://alibaba-damo-academy.github.io/FunASR/en/modelscope_models.html#pretrained-models-on-modelscope), or model path in local disk
-- `ngpu`: 1 (Defalut), decoding on GPU. If ngpu=0, decoding on CPU
-- `ncpu`: 1 (Defalut), sets the number of threads used for intraop parallelism on CPU 
-- `output_dir`: None (Defalut), the output path of results if set
-- `batch_size`: 1 (Defalut), batch size when decoding
-##### infer pipeline
-- `audio_in`: the input to decode, which could be: 
+- `model`: model name in [model zoo](https://alibaba-damo-academy.github.io/FunASR/en/model_zoo/modelscope_models.html#pretrained-models-on-modelscope), or model path in local disk
+- `ngpu`: `1` (Default), decoding on GPU. If ngpu=0, decoding on CPU
+- `ncpu`: `1` (Default), sets the number of threads used for intraop parallelism on CPU
+- `output_dir`: `None` (Default), the output path of results if set
+- `batch_size`: `1` (Default), batch size when decoding
+#### Infer pipeline
+- `audio_in`: the input to decode, which could be:
   - wav_path, `e.g.`: asr_example.wav,
-  - pcm_path, `e.g.`: asr_example.pcm, 
+  - pcm_path, `e.g.`: asr_example.pcm,
   - audio bytes stream, `e.g.`: bytes data from a microphone
   - audio sample point锛宍e.g.`: `audio, rate = soundfile.read("asr_example_zh.wav")`, the dtype is numpy.ndarray or torch.Tensor
-  - wav.scp, kaldi style wav list (`wav_id \t wav_path``), `e.g.`: 
-  ```cat wav.scp
+  - wav.scp, kaldi style wav list (`wav_id \t wav_path`), `e.g.`:
+  ```text
   asr_example1  ./audios/asr_example1.wav
   asr_example2  ./audios/asr_example2.wav
   ```
   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
+- `output_dir`: None (Default), the output path of results if set
 
 ### Inference with multi-thread CPUs or multi GPUs
-FunASR also offer recipes [infer.sh](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/infer.sh) to decode with multi-thread CPUs, or multi GPUs.
+FunASR also offer recipes [egs_modelscope/asr/TEMPLATE/infer.sh](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/infer.sh) to decode with multi-thread CPUs, or multi GPUs.
 
-- 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`
+#### Settings of `infer.sh`
+- `model`: model name in [model zoo](https://alibaba-damo-academy.github.io/FunASR/en/model_zoo/modelscope_models.html#pretrained-models-on-modelscope), or model path in local disk
+- `data_dir`: the dataset dir needs to include `wav.scp`. If `${data_dir}/text` is also exists, CER will be computed
+- `output_dir`: output dir of the recognition results
+- `batch_size`: `64` (Default), batch size of inference on gpu
+- `gpu_inference`: `true` (Default), whether to perform gpu decoding, set false for CPU inference
+- `gpuid_list`: `0,1` (Default), which gpu_ids are used to infer
+- `njob`: only used for CPU inference (`gpu_inference`=`false`), `64` (Default), the number of jobs for CPU decoding
+- `checkpoint_dir`: only used for infer finetuned models, the path dir of finetuned models
+- `checkpoint_name`: only used for infer finetuned models, `valid.cer_ctc.ave.pb` (Default), which checkpoint is used to infer
+- `decoding_mode`: `normal` (Default), decoding mode for UniASR model(fast銆乶ormal銆乷ffline)
+- `hotword_txt`: `None` (Default), hotword file for contextual paraformer model(the hotword file name ends with .txt")
 
-- Decode with multi GPUs:
+#### Decode with multi GPUs:
 ```shell
     bash infer.sh \
     --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
@@ -103,7 +147,7 @@
     --gpu_inference true \
     --gpuid_list "0,1"
 ```
-- Decode with multi-thread CPUs:
+#### Decode with multi-thread CPUs:
 ```shell
     bash infer.sh \
     --model "damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch" \
@@ -113,7 +157,7 @@
     --njob 64
 ```
 
-- Results
+#### 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.
 
@@ -126,15 +170,19 @@
 [finetune.py](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/finetune.py)
 ```python
 import os
+
 from modelscope.metainfo import Trainers
 from modelscope.trainers import build_trainer
-from modelscope.msdatasets.audio.asr_dataset import ASRDataset
+
+from funasr.datasets.ms_dataset import MsDataset
+from funasr.utils.modelscope_param import modelscope_args
+
 
 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')
+    ds_dict = MsDataset.load(params.data_path)
     kwargs = dict(
         model=params.model,
         data_dir=ds_dict,
@@ -142,21 +190,32 @@
         work_dir=params.output_dir,
         batch_bins=params.batch_bins,
         max_epoch=params.max_epoch,
-        lr=params.lr)
+        lr=params.lr,
+        mate_params=params.param_dict)
     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�
-    
+    params = modelscope_args(model="damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch", data_path="./data")
+    params.output_dir = "./checkpoint"              # m妯″瀷淇濆瓨璺緞
+    params.data_path = "./example_data/"            # 鏁版嵁璺緞
+    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 = 20                           # 鏈�澶ц缁冭疆鏁�
+    params.lr = 0.00005                             # 璁剧疆瀛︿範鐜�
+    init_param = []                                 # 鍒濆妯″瀷璺緞锛岄粯璁ゅ姞杞絤odelscope妯″瀷鍒濆鍖栵紝渚嬪: ["checkpoint/20epoch.pb"]
+    freeze_param = []                               # 妯″瀷鍙傛暟freeze, 渚嬪: ["encoder"]
+    ignore_init_mismatch = True                     # 鏄惁蹇界暐妯″瀷鍙傛暟鍒濆鍖栦笉鍖归厤
+    use_lora = False                                # 鏄惁浣跨敤lora杩涜妯″瀷寰皟
+    params.param_dict = {"init_param":init_param, "freeze_param": freeze_param, "ignore_init_mismatch": ignore_init_mismatch}
+    if use_lora:
+        enable_lora = True
+        lora_bias = "all"
+        lora_params = {"lora_list":['q','v'], "lora_rank":8, "lora_alpha":16, "lora_dropout":0.1}
+        lora_config = {"enable_lora": enable_lora, "lora_bias": lora_bias, "lora_params": lora_params}
+        params.param_dict.update(lora_config)
+
     modelscope_finetune(params)
 ```
 
@@ -167,12 +226,31 @@
 ### Finetune with your data
 
 - Modify finetune training related parameters in [finetune.py](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/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
+    - `output_dir`: result dir
+    - `data_dir`: the dataset dir needs to include files: `train/wav.scp`, `train/text`; `validation/wav.scp`, `validation/text`
+    - `dataset_type`: for dataset larger than 1000 hours, set as `large`, otherwise set as `small`
+    - `batch_bins`: 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
+    - `max_epoch`: number of training epoch
+    - `lr`: learning rate
+    - `init_param`: `[]`(Default), init model path, load modelscope model initialization by default. For example: ["checkpoint/20epoch.pb"]
+    - `freeze_param`: `[]`(Default), Freeze model parameters. For example锛歔"encoder"]
+    - `ignore_init_mismatch`: `True`(Default), Ignore size mismatch when loading pre-trained model
+    - `use_lora`: `False`(Default), Fine-tuning model use lora, more detail please refer to [LORA](https://arxiv.org/pdf/2106.09685.pdf)
+
+- Training data formats锛�
+```sh
+cat ./example_data/text
+BAC009S0002W0122 鑰� 瀵� 妤� 甯� 鎴� 浜� 鎶� 鍒� 浣� 鐢� 鏈� 澶� 鐨� 闄� 璐�
+BAC009S0002W0123 涔� 鎴� 涓� 鍦� 鏂� 鏀� 搴� 鐨� 鐪� 涓� 閽�
+english_example_1 hello world
+english_example_2 go swim 鍘� 娓� 娉�
+
+cat ./example_data/wav.scp
+BAC009S0002W0122 /mnt/data/wav/train/S0002/BAC009S0002W0122.wav
+BAC009S0002W0123 /mnt/data/wav/train/S0002/BAC009S0002W0123.wav
+english_example_1 /mnt/data/wav/train/S0002/english_example_1.wav
+english_example_2 /mnt/data/wav/train/S0002/english_example_2.wav
+```
 
 - Then you can run the pipeline to finetune with:
 ```shell
@@ -183,14 +261,29 @@
 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](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/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
+- Setting parameters in [egs_modelscope/asr/TEMPLATE/infer.sh](https://github.com/alibaba-damo-academy/FunASR/blob/main/egs_modelscope/asr/TEMPLATE/infer.sh) is the same with [docs](https://github.com/alibaba-damo-academy/FunASR/tree/main/egs_modelscope/asr/TEMPLATE#inference-with-multi-thread-cpus-or-multi-gpus), `model` is the model name from modelscope, which you finetuned.
+
+- 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" \
+    --checkpoint_dir "./checkpoint" \
+    --checkpoint_name "valid.cer_ctc.ave.pb"
+```
+- 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 \
+    --checkpoint_dir "./checkpoint" \
+    --checkpoint_name "valid.cer_ctc.ave.pb"
 ```

--
Gitblit v1.9.1