mirror of
https://github.com/modelscope/FunASR
synced 2025-09-15 14:48:36 +08:00
tutorial
This commit is contained in:
游雁
parent
7a4d17b54d
commit
944a053a66
@ -1,267 +0,0 @@
|
||||
# Speech Recognition
|
||||
In FunASR, we provide several ASR benchmarks, such as AISHLL, Librispeech, WenetSpeech, while different model architectures are supported, including conformer, paraformer, uniasr.
|
||||
|
||||
## Quick Start
|
||||
After downloaded and installed FunASR, users can use our provided recipes to easily reproduce the relevant experimental results. Here we take "paraformer on AISHELL-1" as an example.
|
||||
|
||||
First, move to the corresponding dictionary of the AISHELL-1 paraformer example.
|
||||
```sh
|
||||
cd egs/aishell/paraformer
|
||||
```
|
||||
|
||||
Then you can directly start the recipe as follows:
|
||||
```sh
|
||||
conda activate funasr
|
||||
bash run.sh --CUDA_VISIBLE_DEVICES "0,1" --gpu_num 2
|
||||
```
|
||||
|
||||
The training log files are saved in `${exp_dir}/exp/${model_dir}/log/train.log.*`， which can be viewed using the following command:
|
||||
```sh
|
||||
vim exp/*_train_*/log/train.log.0
|
||||
```
|
||||
|
||||
Users can observe the training loss, prediction accuracy and other training information, like follows:
|
||||
```text
|
||||
... 1epoch:train:751-800batch:800num_updates: ... loss_ctc=106.703, loss_att=86.877, acc=0.029, loss_pre=1.552 ...
|
||||
... 1epoch:train:801-850batch:850num_updates: ... loss_ctc=107.890, loss_att=87.832, acc=0.029, loss_pre=1.702 ...
|
||||
```
|
||||
|
||||
At the end of each epoch, the evaluation metrics are calculated on the validation set, like follows:
|
||||
```text
|
||||
... [valid] loss_ctc=99.914, cer_ctc=1.000, loss_att=80.512, acc=0.029, cer=0.971, wer=1.000, loss_pre=1.952, loss=88.285 ...
|
||||
```
|
||||
|
||||
Also, users can use tensorboard to observe these training information by the following command:
|
||||
```sh
|
||||
tensorboard --logdir ${exp_dir}/exp/${model_dir}/tensorboard/train
|
||||
```
|
||||
Here is an example of loss:
|
||||
|
||||
<img src="./academic_recipe/images/loss.png" width="200"/>
|
||||
|
||||
The inference results are saved in `${exp_dir}/exp/${model_dir}/decode_asr_*/$dset`. The main two files are `text.cer` and `text.cer.txt`. `text.cer` saves the comparison between the recognized text and the reference text, like follows:
|
||||
```text
|
||||
...
|
||||
BAC009S0764W0213(nwords=11,cor=11,ins=0,del=0,sub=0) corr=100.00%,cer=0.00%
|
||||
ref: 构 建 良 好 的 旅 游 市 场 环 境
|
||||
res: 构 建 良 好 的 旅 游 市 场 环 境
|
||||
...
|
||||
```
|
||||
`text.cer.txt` saves the final results, like follows:
|
||||
```text
|
||||
%WER ...
|
||||
%SER ...
|
||||
Scored ... sentences, ...
|
||||
```
|
||||
|
||||
## Introduction
|
||||
We provide a recipe `egs/aishell/paraformer/run.sh` for training a paraformer model on AISHELL-1 dataset. This recipe consists of five stages, supporting training on multiple GPUs and decoding by CPU or GPU. Before introducing each stage in detail, we first explain several parameters which should be set by users.
|
||||
- `CUDA_VISIBLE_DEVICES`: `0,1` (Default), visible gpu list
|
||||
- `gpu_num`: `2` (Default), the number of GPUs used for training
|
||||
- `gpu_inference`: `true` (Default), whether to use GPUs for decoding
|
||||
- `njob`: `1` (Default),for CPU decoding, indicating the total number of CPU jobs; for GPU decoding, indicating the number of jobs on each GPU
|
||||
- `raw_data`: the raw path of AISHELL-1 dataset
|
||||
- `feats_dir`: the path for saving processed data
|
||||
- `token_type`: `char` (Default), indicate how to process text
|
||||
- `type`: `sound` (Default), set the input type
|
||||
- `scp`: `wav.scp` (Default), set the input file
|
||||
- `nj`: `64` (Default), the number of jobs for data preparation
|
||||
- `speed_perturb`: `"0.9, 1.0 ,1.1"` (Default), the range of speech perturbed
|
||||
- `exp_dir`: the path for saving experimental results
|
||||
- `tag`: `exp1` (Default), the suffix of experimental result directory
|
||||
- `stage` `0` (Default), start the recipe from the specified stage
|
||||
- `stop_stage` `5` (Default), stop the recipe from the specified stage
|
||||
|
||||
### Stage 0: Data preparation
|
||||
This stage processes raw AISHELL-1 dataset `$raw_data` and generates the corresponding `wav.scp` and `text` in `$feats_dir/data/xxx`. `xxx` means `train/dev/test`. Here we assume users have already downloaded AISHELL-1 dataset. If not, users can download data [here](https://www.openslr.org/33/) and set the path for `$raw_data`. The examples of `wav.scp` and `text` are as follows:
|
||||
* `wav.scp`
|
||||
```
|
||||
BAC009S0002W0122 /nfs/ASR_DATA/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0122.wav
|
||||
BAC009S0002W0123 /nfs/ASR_DATA/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0123.wav
|
||||
BAC009S0002W0124 /nfs/ASR_DATA/AISHELL-1/data_aishell/wav/train/S0002/BAC009S0002W0124.wav
|
||||
...
|
||||
```
|
||||
* `text`
|
||||
```
|
||||
BAC009S0002W0122 而 对 楼 市 成 交 抑 制 作 用 最 大 的 限 购
|
||||
BAC009S0002W0123 也 成 为 地 方 政 府 的 眼 中 钉
|
||||
BAC009S0002W0124 自 六 月 底 呼 和 浩 特 市 率 先 宣 布 取 消 限 购 后
|
||||
...
|
||||
```
|
||||
These two files both have two columns, while the first column is wav ids and the second column is the corresponding wav paths/label tokens.
|
||||
|
||||
### Stage 1: Feature and CMVN Generation
|
||||
This stage computes CMVN based on `train` dataset, which is used in the following stages. Users can set `nj` to control the number of jobs for computing CMVN. The generated CMVN file is saved as `$feats_dir/data/train/cmvn/am.mvn`.
|
||||
|
||||
### Stage 2: Dictionary Preparation
|
||||
This stage processes the dictionary, which is used as a mapping between label characters and integer indices during ASR training. The processed dictionary file is saved as `$feats_dir/data/$lang_toekn_list/$token_type/tokens.txt`. An example of `tokens.txt` is as follows:
|
||||
```
|
||||
<blank>
|
||||
<s>
|
||||
</s>
|
||||
一
|
||||
丁
|
||||
...
|
||||
龚
|
||||
龟
|
||||
<unk>
|
||||
```
|
||||
There are four tokens must be specified:
|
||||
* `<blank>`: (required), indicates the blank token for CTC, must be in the first line
|
||||
* `<s>`: (required), indicates the start-of-sentence token, must be in the second line
|
||||
* `</s>`: (required), indicates the end-of-sentence token, must be in the third line
|
||||
* `<unk>`: (required), indicates the out-of-vocabulary token, must be in the last line
|
||||
|
||||
### Stage 3: LM Training
|
||||
|
||||
### Stage 4: ASR Training
|
||||
This stage achieves the training of the specified model. To start training, users should manually set `exp_dir` to specify the path for saving experimental results. By default, the best `$keep_nbest_models` checkpoints on validation dataset will be averaged to generate a better model and adopted for decoding. FunASR implements `train.py` for training different models and users can configure the following parameters if necessary. The training command is as follows:
|
||||
|
||||
```sh
|
||||
train.py \
|
||||
--task_name asr \
|
||||
--use_preprocessor true \
|
||||
--token_list $token_list \
|
||||
--data_dir ${feats_dir}/data \
|
||||
--train_set ${train_set} \
|
||||
--valid_set ${valid_set} \
|
||||
--data_file_names "wav.scp,text" \
|
||||
--cmvn_file ${feats_dir}/data/${train_set}/cmvn/am.mvn \
|
||||
--speed_perturb ${speed_perturb} \
|
||||
--resume true \
|
||||
--output_dir ${exp_dir}/exp/${model_dir} \
|
||||
--config $asr_config \
|
||||
--ngpu $gpu_num \
|
||||
...
|
||||
```
|
||||
|
||||
* `task_name`: `asr` (Default), specify the task type of the current recipe
|
||||
* `ngpu`: `2` (Default), specify the number of GPUs for training. When `ngpu > 1`, DistributedDataParallel (DDP, the detail can be found [here](https://pytorch.org/tutorials/intermediate/ddp_tutorial.html)) training will be enabled. Correspondingly, `CUDA_VISIBLE_DEVICES` should be set to specify which ids of GPUs will be used.
|
||||
* `use_preprocessor`: `true` (Default), specify whether to use pre-processing on each sample
|
||||
* `token_list`: the path of token list for training
|
||||
* `dataset_type`: `small` (Default). FunASR supports `small` dataset type for training small datasets. Besides, an optional iterable-style DataLoader based on [Pytorch Iterable-style DataPipes](https://pytorch.org/data/beta/torchdata.datapipes.iter.html) for large datasets is supported and users can specify `dataset_type=large` to enable it.
|
||||
* `data_dir`: the path of data. Specifically, the data for training is saved in `$data_dir/data/$train_set` while the data for validation is saved in `$data_dir/data/$valid_set`
|
||||
* `data_file_names`: `"wav.scp,text"` specify the speech and text file names for ASR
|
||||
* `cmvn_file`: the path of cmvn file
|
||||
* `resume`: `true`, whether to enable "checkpoint training"
|
||||
* `output_dir`: the path for saving training results
|
||||
* `config`: the path of configuration file, which is usually a YAML file in `conf` directory. In FunASR, the parameters of the training, including model, optimization, dataset, etc., can also be set in this file. Note that if the same parameters are specified in both recipe and config file, the parameters of recipe will be employed
|
||||
|
||||
### Stage 5: Decoding
|
||||
This stage generates the recognition results and calculates the `CER` to verify the performance of the trained model.
|
||||
|
||||
* Mode Selection
|
||||
|
||||
As we support paraformer, uniasr, conformer and other models in FunASR, a `mode` parameter should be specified as `asr/paraformer/uniasr` according to the trained model.
|
||||
|
||||
* Configuration
|
||||
|
||||
We support CTC decoding, attention decoding and hybrid CTC-attention decoding in FunASR, which can be specified by `ctc_weight` in a YAML file in `conf` directory. Specifically, `ctc_weight=1.0` indicates CTC decoding, `ctc_weight=0.0` indicates attention decoding, `0.0<ctc_weight<1.0` indicates hybrid CTC-attention decoding.
|
||||
|
||||
* CPU/GPU Decoding
|
||||
|
||||
We support CPU and GPU decoding in FunASR. For CPU decoding, you should set `gpu_inference=False` and set `njob` to specify the total number of CPU decoding jobs. For GPU decoding, you should set `gpu_inference=True`. You should also set `gpuid_list` to indicate which GPUs are used for decoding and `njobs` to indicate the number of decoding jobs on each GPU.
|
||||
|
||||
* Performance
|
||||
|
||||
We adopt `CER` to verify the performance. The results are in `$exp_dir/exp/$model_dir/$decoding_yaml_name/$average_model_name/$dset`, namely `text.cer` and `text.cer.txt`. `text.cer` saves the comparison between the recognized text and the reference text while `text.cer.txt` saves the final `CER` results. The following is an example of `text.cer`:
|
||||
```
|
||||
...
|
||||
BAC009S0764W0213(nwords=11,cor=11,ins=0,del=0,sub=0) corr=100.00%,cer=0.00%
|
||||
ref: 构 建 良 好 的 旅 游 市 场 环 境
|
||||
res: 构 建 良 好 的 旅 游 市 场 环 境
|
||||
...
|
||||
```
|
||||
|
||||
## Change settings
|
||||
Here we explain how to perform common custom settings, which can help users to modify scripts according to their own needs.
|
||||
|
||||
### Training with specified GPUs
|
||||
|
||||
For example, if users want to use 2 GPUs with id `2` and `3`, users can run the following command:
|
||||
```sh
|
||||
. ./run.sh --CUDA_VISIBLE_DEVICES "2,3" --gpu_num 2
|
||||
```
|
||||
|
||||
### Start from/Stop at a specified stage
|
||||
|
||||
The recipe includes several stages. Users can start form or stop at any stage. For example, the following command achieves starting from the third stage and stopping at the fifth stage:
|
||||
```sh
|
||||
. ./run.sh --stage 3 --stop_stage 5
|
||||
```
|
||||
|
||||
### Specify total training steps
|
||||
|
||||
FunASR supports two parameters to specify the training steps, namely `max_epoch` and `max_update`. `max_epoch` indicates the total training epochs while `max_update` indicates the total training steps. If these two parameters are specified at the same time, once the training reaches any one of these two parameters, the training will be stopped.
|
||||
|
||||
### Change the configuration of the model
|
||||
|
||||
The configuration of the model is set in the config file `conf/train_*.yaml`. Specifically, the default encoder configuration of paraformer is as follows:
|
||||
```
|
||||
encoder: conformer
|
||||
encoder_conf:
|
||||
output_size: 256 # dimension of attention
|
||||
attention_heads: 4 # the number of heads in multi-head attention
|
||||
linear_units: 2048 # the number of units of position-wise feed forward
|
||||
num_blocks: 12 # the number of encoder blocks
|
||||
dropout_rate: 0.1
|
||||
positional_dropout_rate: 0.1
|
||||
attention_dropout_rate: 0.0
|
||||
input_layer: conv2d # encoder input layer architecture type
|
||||
normalize_before: true
|
||||
pos_enc_layer_type: rel_pos
|
||||
selfattention_layer_type: rel_selfattn
|
||||
activation_type: swish
|
||||
macaron_style: true
|
||||
use_cnn_module: true
|
||||
cnn_module_kernel: 15
|
||||
|
||||
```
|
||||
Users can change the encoder configuration by modify these values. For example, if users want to use an encoder with 16 conformer blocks and each block has 8 attention heads, users just need to change `num_blocks` from 12 to 16 and change `attention_heads` from 4 to 8. Besides, the batch_size, learning rate and other training hyper-parameters are also set in this config file. To change these hyper-parameters, users just need to directly change the corresponding values in this file. For example, the default learning rate is `0.0005`. If users want to change the learning rate to 0.0002, set the value of lr as `lr: 0.0002`.
|
||||
|
||||
### Change different input data type
|
||||
|
||||
FunASR supports different input data types, including `sound`, `kaldi_ark`, `npy`, `text` and `text_int`. Users can specify any number and any type of input, which is achieved by `data_names` and `data_types` (in `config/train_*.yaml`). For example, ASR task usually requires speech and the transcripts as input. In FunASR, by default, speech is saved as raw audio (such as wav format) and transcripts are saved as text format. Correspondingly, `data_names` and `data_types` are set as follows (seen in `config/train_*.yaml`):
|
||||
```text
|
||||
dataset_conf:
|
||||
data_names: speech,text
|
||||
data_types: sound,text
|
||||
...
|
||||
```
|
||||
When the input type changes to FBank, users just need to modify as `data_types: kaldi_ark,text` in the config file. Note `data_file_names` used in `train.py` should also be changed to the new file name.
|
||||
|
||||
### How to resume training process
|
||||
FunASR supports resuming training as follows:
|
||||
```shell
|
||||
train.py ... --resume true ...
|
||||
```
|
||||
|
||||
### How to transfer / fine-tuning from pre-trained models
|
||||
|
||||
FunASR supports transferring / fine-tuning from a pre-trained model by specifying the `init_param` parameter. The usage format is as follows:
|
||||
```shell
|
||||
train.py ... --init_param <file_path>:<src_key>:<dst_key>:<exclude_keys> ..
|
||||
```
|
||||
For example, the following command achieves loading all pretrained parameters starting from decoder except decoder.embed and set it to model.decoder2:
|
||||
```shell
|
||||
train.py ... --init_param model.pb:decoder:decoder2:decoder.embed ...
|
||||
```
|
||||
Besides, loading parameters from multiple pre-trained models is supported. For example, the following command achieves loading encoder parameters from the pre-trained model1 and decoder parameters from the pre-trained model2:
|
||||
```sh
|
||||
train.py ... --init_param model1.pb:encoder --init_param model2.pb:decoder ...
|
||||
```
|
||||
|
||||
### How to freeze part of the model parameters
|
||||
|
||||
In certain situations, users may want to fix part of the model parameters update the rest model parameters. FunASR employs `freeze_param` to achieve this. For example, to fix all parameters like `encoder.*`, users need to set `freeze_param ` as follows:
|
||||
```sh
|
||||
train.py ... --freeze_param encoder ...
|
||||
```
|
||||
|
||||
### ModelScope Usage
|
||||
|
||||
Users can use ModelScope for inference and fine-tuning based on a trained academic model. To achieve this, users need to run the stage 6 in the script. In this stage, relevant files required by ModelScope will be generated automatically. Users can then use the corresponding ModelScope interface by replacing the model name with the local trained model path. For the detailed usage of the ModelScope interface, please refer to [ModelScope Usage](https://alibaba-damo-academy.github.io/FunASR/en/modelscope_pipeline/quick_start.html).
|
||||
|
||||
### Decoding by CPU or GPU
|
||||
|
||||
We support CPU and GPU decoding. For CPU decoding, set `gpu_inference=false` and `njob` to specific the total number of CPU jobs. For GPU decoding, first set `gpu_inference=true`. Then set `gpuid_list` to specific which GPUs for decoding and `njob` to specific the number of decoding jobs on each GPU.
|
||||
Binary file not shown.
|
Before Width: | Height: | Size: 528 KiB |
@ -1,3 +0,0 @@
|
||||
# Speech Recognition
|
||||
|
||||
Undo
|
||||
@ -1,2 +0,0 @@
|
||||
# Punctuation Restoration
|
||||
Undo
|
||||
@ -1,2 +0,0 @@
|
||||
# Speaker Diarization
|
||||
Undo
|
||||
@ -1,2 +0,0 @@
|
||||
# Speaker Verification
|
||||
Undo
|
||||
@ -1,2 +0,0 @@
|
||||
# Voice Activity Detection
|
||||
Undo
|
||||
@ -1 +0,0 @@
|
||||
../funasr
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/asr/TEMPLATE/README.md
|
||||
@ -1,63 +0,0 @@
|
||||
# Inverse Text Normalization (ITN)
|
||||
|
||||
> **Note**:
|
||||
> The modelscope pipeline supports all the models in [model zoo](https://modelscope.cn/models?page=1&tasks=inverse-text-processing&type=audio) to inference. Here we take the model of the Japanese ITN model as example to demonstrate the usage.
|
||||
|
||||
## Inference
|
||||
|
||||
### Quick start
|
||||
#### [Japanese ITN model](https://modelscope.cn/models/damo/speech_inverse_text_processing_fun-text-processing-itn-ja/summary)
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
itn_inference_pipline = pipeline(
|
||||
task=Tasks.inverse_text_processing,
|
||||
model='damo/speech_inverse_text_processing_fun-text-processing-itn-ja',
|
||||
model_revision=None)
|
||||
|
||||
itn_result = itn_inference_pipline(text_in='百二十三')
|
||||
print(itn_result)
|
||||
# 123
|
||||
```
|
||||
- read text data directly.
|
||||
```python
|
||||
rec_result = inference_pipeline(text_in='一九九九年に誕生した同商品にちなみ、約三十年前、二十四歳の頃の幸四郎の写真を公開。')
|
||||
# 1999年に誕生した同商品にちなみ、約30年前、24歳の頃の幸四郎の写真を公開。
|
||||
```
|
||||
- text stored via url，example：https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/ja_itn_example.txt
|
||||
```python
|
||||
rec_result = inference_pipeline(text_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/ja_itn_example.txt')
|
||||
```
|
||||
|
||||
Full code of demo, please ref to [demo](https://github.com/alibaba-damo-academy/FunASR/tree/main/fun_text_processing/inverse_text_normalization)
|
||||
|
||||
### API-reference
|
||||
#### Define pipeline
|
||||
- `task`: `Tasks.inverse_text_processing`
|
||||
- `model`: model name in [model zoo](https://modelscope.cn/models?page=1&tasks=inverse-text-processing&type=audio), or model path in local disk
|
||||
- `output_dir`: `None` (Default), the output path of results if set
|
||||
- `model_revision`: `None` (Default), setting the model version
|
||||
|
||||
#### Infer pipeline
|
||||
- `text_in`: the input to decode, which could be:
|
||||
- text bytes, `e.g.`: "一九九九年に誕生した同商品にちなみ、約三十年前、二十四歳の頃の幸四郎の写真を公開。"
|
||||
- text file, `e.g.`: https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/ja_itn_example.txt
|
||||
In this case of `text file` input, `output_dir` must be set to save the output results
|
||||
|
||||
## Modify Your Own ITN Model
|
||||
The rule-based ITN code is open-sourced in [FunTextProcessing](https://github.com/alibaba-damo-academy/FunASR/tree/main/fun_text_processing), users can modify by their own grammar rules for different languages. Let's take Japanese as an example, users can add their own whitelist in ```FunASR/fun_text_processing/inverse_text_normalization/ja/data/whitelist.tsv```. After modified the grammar rules, the users can export and evaluate their own ITN models in local directory.
|
||||
|
||||
### Export ITN Model
|
||||
Export ITN model via ```FunASR/fun_text_processing/inverse_text_normalization/export_models.py```. An example to export ITN model to local folder is shown as below.
|
||||
```shell
|
||||
cd FunASR/fun_text_processing/inverse_text_normalization/
|
||||
python export_models.py --language ja --export_dir ./itn_models/
|
||||
```
|
||||
|
||||
### Evaluate ITN Model
|
||||
Users can evaluate their own ITN model in local directory via ```FunASR/fun_text_processing/inverse_text_normalization/inverse_normalize.py```. Here is an example:
|
||||
```shell
|
||||
cd FunASR/fun_text_processing/inverse_text_normalization/
|
||||
python inverse_normalize.py --input_file ja_itn_example.txt --cache_dir ./itn_models/ --output_file output.txt --language=ja
|
||||
```
|
||||
@ -1,14 +0,0 @@
|
||||
# Language Models
|
||||
|
||||
## Inference with pipeline
|
||||
### Quick start
|
||||
### Inference with you data
|
||||
### Inference with multi-threads on CPU
|
||||
### Inference with multi GPU
|
||||
|
||||
## Finetune with pipeline
|
||||
### Quick start
|
||||
### Finetune with your data
|
||||
|
||||
## Inference with your finetuned model
|
||||
|
||||
@ -1,53 +0,0 @@
|
||||
# ModelScope Usage
|
||||
ModelScope is an open-source model-as-service platform supported by Alibaba, which provides flexible and convenient model applications for users in academia and industry. For specific usages and open source models, please refer to [ModelScope](https://www.modelscope.cn/models?page=1&tasks=auto-speech-recognition). In the domain of speech, we provide autoregressive/non-autoregressive speech recognition, speech pre-training, punctuation prediction and other models, which are convenient for users.
|
||||
|
||||
## Overall Introduction
|
||||
We provide the usages of different models under the `egs_modelscope`, which supports directly employing our provided models for inference, as well as finetuning the models we provided as pre-trained initial models. Next, we will introduce the model provided in the `egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch` directory, including `infer.py`, `finetune.py` and `infer_after_finetune .py`. The corresponding functions are as follows:
|
||||
- `infer.py`: perform inference on the specified dataset based on our provided model
|
||||
- `finetune.py`: employ our provided model as the initial model for fintuning
|
||||
- `infer_after_finetune.py`: perform inference on the specified dataset based on the finetuned model
|
||||
|
||||
## Inference
|
||||
We provide `infer.py` to achieve the inference. Based on this file, users can preform inference on the specified dataset based on our provided model and obtain the corresponding recognition results. If the transcript is given, the `CER` will be calculated at the same time. Before performing inference, users can set the following parameters to modify the inference configuration:
|
||||
* `data_dir`：dataset directory. The directory should contain the wav list file `wav.scp` and the transcript file `text` (optional). For the format of these two files, please refer to the instructions in [Quick Start](./get_started.md). If the `text` file exists, the CER will be calculated accordingly, otherwise it will be skipped.
|
||||
* `output_dir`：the directory for saving the inference results
|
||||
* `batch_size`：batch size during the inference
|
||||
* `ctc_weight`：some models contain a CTC module, users can set this parameter to specify the weight of the CTC module during the inference
|
||||
|
||||
In addition to directly setting parameters in `infer.py`, users can also manually set the parameters in the `decoding.yaml` file in the model download directory to modify the inference configuration.
|
||||
|
||||
## Finetuning
|
||||
We provide `finetune.py` to achieve the finetuning. Based on this file, users can finetune on the specified dataset based on our provided model as the initial model to achieve better performance in the specificed domain. Before finetuning, users can set the following parameters to modify the finetuning configuration:
|
||||
* `data_path`：dataset directory。This directory should contain the `train` directory for saving the training set and the `dev` directory for saving the validation set. Each directory needs to contain the wav list file `wav.scp` and the transcript file `text`
|
||||
* `output_dir`：the directory for saving the finetuning results
|
||||
* `dataset_type`：for small dataset，set as `small`；for dataset larger than 1000 hours，set as `large`
|
||||
* `batch_bins`：batch size，if dataset_type is set as `small`，the unit of batch_bins is the number of fbank feature frames; if dataset_type is set as `large`, the unit of batch_bins is milliseconds
|
||||
* `max_epoch`：the maximum number of training epochs
|
||||
|
||||
The following parameters can also be set. However, if there is no special requirement, users can ignore these parameters and use the default value we provided directly:
|
||||
* `accum_grad`：the accumulation of the gradient
|
||||
* `keep_nbest_models`：select the `keep_nbest_models` models with the best performance and average the parameters
|
||||
of these models to get a better model
|
||||
* `optim`：set the optimizer
|
||||
* `lr`：set the learning rate
|
||||
* `scheduler`：set learning rate adjustment strategy
|
||||
* `scheduler_conf`：set the related parameters of the learning rate adjustment strategy
|
||||
* `specaug`：set for the spectral augmentation
|
||||
* `specaug_conf`：set related parameters of the spectral augmentation
|
||||
|
||||
In addition to directly setting parameters in `finetune.py`, users can also manually set the parameters in the `finetune.yaml` file in the model download directory to modify the finetuning configuration.
|
||||
|
||||
## Inference after Finetuning
|
||||
We provide `infer_after_finetune.py` to achieve the inference based on the model finetuned by users. Based on this file, users can preform inference on the specified dataset based on the finetuned model and obtain the corresponding recognition results. If the transcript is given, the `CER` will be calculated at the same time. Before performing inference, users can set the following parameters to modify the inference configuration:
|
||||
* `data_dir`：dataset directory。The directory should contain the wav list file `wav.scp` and the transcript file `text` (optional). If the `text` file exists, the CER will be calculated accordingly, otherwise it will be skipped.
|
||||
* `output_dir`：the directory for saving the inference results
|
||||
* `batch_size`：batch size during the inference
|
||||
* `ctc_weight`：some models contain a CTC module, users can set this parameter to specify the weight of the CTC module during the inference
|
||||
* `decoding_model_name`：set the name of the model used for the inference
|
||||
|
||||
The following parameters can also be set. However, if there is no special requirement, users can ignore these parameters and use the default value we provided directly:
|
||||
* `modelscope_model_name`：the initial model name used when finetuning
|
||||
* `required_files`：files required for the inference when using the modelscope interface
|
||||
|
||||
## Announcements
|
||||
Some models may have other specific parameters during the finetuning and inference. The usages of these parameters can be found in the `README.md` file in the corresponding directory.
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/punctuation/TEMPLATE/README.md
|
||||
@ -1,226 +0,0 @@
|
||||
([简体中文](./quick_start_zh.md)|English)
|
||||
|
||||
# Quick Start
|
||||
|
||||
> **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 typic model as example to demonstrate the usage.
|
||||
|
||||
|
||||
## Inference with pipeline
|
||||
|
||||
### Speech Recognition
|
||||
#### Paraformer Model
|
||||
```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-pytorch',
|
||||
)
|
||||
|
||||
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)
|
||||
# {'text': '欢迎大家来体验达摩院推出的语音识别模型'}
|
||||
```
|
||||
|
||||
### Voice Activity Detection
|
||||
#### FSMN-VAD Model
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
from modelscope.utils.logger import get_logger
|
||||
import logging
|
||||
logger = get_logger(log_level=logging.CRITICAL)
|
||||
logger.setLevel(logging.CRITICAL)
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.voice_activity_detection,
|
||||
model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch',
|
||||
)
|
||||
|
||||
segments_result = inference_pipeline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav')
|
||||
print(segments_result)
|
||||
# {'text': [[70, 2340], [2620, 6200], [6480, 23670], [23950, 26250], [26780, 28990], [29950, 31430], [31750, 37600], [38210, 46900], [47310, 49630], [49910, 56460], [56740, 59540], [59820, 70450]]}
|
||||
```
|
||||
|
||||
### Punctuation Restoration
|
||||
#### CT_Transformer Model
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.punctuation,
|
||||
model='damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
|
||||
)
|
||||
|
||||
rec_result = inference_pipeline(text_in='我们都是木头人不会讲话不会动')
|
||||
print(rec_result)
|
||||
# {'text': '我们都是木头人，不会讲话，不会动。'}
|
||||
```
|
||||
|
||||
### Timestamp Prediction
|
||||
#### TP-Aligner Model
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.speech_timestamp,
|
||||
model='damo/speech_timestamp_prediction-v1-16k-offline',)
|
||||
|
||||
rec_result = inference_pipeline(
|
||||
audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_timestamps.wav',
|
||||
text_in='一 个 东 太 平 洋 国 家 为 什 么 跑 到 西 太 平 洋 来 了 呢',)
|
||||
print(rec_result)
|
||||
# {'text': '<sil> 0.000 0.380;一 0.380 0.560;个 0.560 0.800;东 0.800 0.980;太 0.980 1.140;平 1.140 1.260;洋 1.260 1.440;国 1.440 1.680;家 1.680 1.920;<sil> 1.920 2.040;为 2.040 2.200;什 2.200 2.320;么 2.320 2.500;跑 2.500 2.680;到 2.680 2.860;西 2.860 3.040;太 3.040 3.200;平 3.200 3.380;洋 3.380 3.500;来 3.500 3.640;了 3.640 3.800;呢 3.800 4.150;<sil> 4.150 4.440;', 'timestamp': [[380, 560], [560, 800], [800, 980], [980, 1140], [1140, 1260], [1260, 1440], [1440, 1680], [1680, 1920], [2040, 2200], [2200, 2320], [2320, 2500], [2500, 2680], [2680, 2860], [2860, 3040], [3040, 3200], [3200, 3380], [3380, 3500], [3500, 3640], [3640, 3800], [3800, 4150]]}
|
||||
```
|
||||
|
||||
### Speaker Verification
|
||||
#### X-vector Model
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
import numpy as np
|
||||
|
||||
inference_sv_pipline = pipeline(
|
||||
task=Tasks.speaker_verification,
|
||||
model='damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch'
|
||||
)
|
||||
|
||||
# embedding extract
|
||||
spk_embedding = inference_sv_pipline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav')["spk_embedding"]
|
||||
|
||||
# speaker verification
|
||||
rec_result = inference_sv_pipline(audio_in=('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav','https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_same.wav'))
|
||||
print(rec_result["scores"][0])
|
||||
# 0.8540499500025098
|
||||
```
|
||||
|
||||
### Speaker Diarization
|
||||
#### SOND Model
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_diar_pipline = pipeline(
|
||||
mode="sond_demo",
|
||||
num_workers=0,
|
||||
task=Tasks.speaker_diarization,
|
||||
diar_model_config="sond.yaml",
|
||||
model='damo/speech_diarization_sond-en-us-callhome-8k-n16k4-pytorch',
|
||||
model_revision="v1.0.3",
|
||||
sv_model="damo/speech_xvector_sv-en-us-callhome-8k-spk6135-pytorch",
|
||||
sv_model_revision="v1.0.0",
|
||||
)
|
||||
|
||||
audio_list=[
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/record.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_A.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_B.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_B1.wav"
|
||||
]
|
||||
|
||||
results = inference_diar_pipline(audio_in=audio_list)
|
||||
print(results)
|
||||
# {'text': 'spk1 [(0.8, 1.84), (2.8, 6.16), (7.04, 10.64), (12.08, 12.8), (14.24, 15.6)]\nspk2 [(0.0, 1.12), (1.68, 3.2), (4.48, 7.12), (8.48, 9.04), (10.56, 14.48), (15.44, 16.0)]'}
|
||||
```
|
||||
|
||||
### FAQ
|
||||
#### How to switch device from GPU to CPU with pipeline
|
||||
|
||||
The pipeline defaults to decoding with GPU (`ngpu=1`) when GPU is available. If you want to switch to CPU, you could set `ngpu=0`
|
||||
```python
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.auto_speech_recognition,
|
||||
model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
|
||||
ngpu=0,
|
||||
)
|
||||
```
|
||||
|
||||
#### How to infer from local model path
|
||||
Download model to local dir, by modelscope-sdk
|
||||
|
||||
```python
|
||||
from modelscope.hub.snapshot_download import snapshot_download
|
||||
|
||||
local_dir_root = "./models_from_modelscope"
|
||||
model_dir = snapshot_download('damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', cache_dir=local_dir_root)
|
||||
```
|
||||
|
||||
Or download model to local dir, by git lfs
|
||||
```shell
|
||||
git lfs install
|
||||
# git clone https://www.modelscope.cn/<namespace>/<model-name>.git
|
||||
git clone https://www.modelscope.cn/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch.git
|
||||
```
|
||||
|
||||
Infer with local model path
|
||||
```python
|
||||
local_dir_root = "./models_from_modelscope/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.auto_speech_recognition,
|
||||
model=local_dir_root,
|
||||
)
|
||||
```
|
||||
|
||||
## Finetune with pipeline
|
||||
### Speech Recognition
|
||||
#### Paraformer Model
|
||||
|
||||
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" # 模型保存路径
|
||||
params.data_path = "speech_asr_aishell1_trainsets" # 数据路径，可以为modelscope中已上传数据，也可以是本地数据
|
||||
params.dataset_type = "small" # 小数据量设置small，若数据量大于1000小时，请使用large
|
||||
params.batch_bins = 2000 # batch size，如果dataset_type="small"，batch_bins单位为fbank特征帧数，如果dataset_type="large"，batch_bins单位为毫秒，
|
||||
params.max_epoch = 50 # 最大训练轮数
|
||||
params.lr = 0.00005 # 设置学习率
|
||||
|
||||
modelscope_finetune(params)
|
||||
```
|
||||
|
||||
```shell
|
||||
python finetune.py &> log.txt &
|
||||
```
|
||||
tail log.txt
|
||||
```
|
||||
[bach-gpu011024008134] 2023-04-23 18:59:13,976 (e2e_asr_paraformer:467) INFO: enable sampler in paraformer, sampling_ratio: 0.75
|
||||
[bach-gpu011024008134] 2023-04-23 18:59:48,924 (trainer:777) INFO: 2epoch:train:1-50batch:50num_updates: iter_time=0.008, forward_time=0.302, loss_att=0.186, acc=0.942, loss_pre=0.005, loss=0.192, backward_time=0.231, optim_step_time=0.117, optim0_lr0=7.484e-06, train_time=0.753
|
||||
[bach-gpu011024008134] 2023-04-23 19:00:23,869 (trainer:777) INFO: 2epoch:train:51-100batch:100num_updates: iter_time=1.152e-04, forward_time=0.275, loss_att=0.184, acc=0.945, loss_pre=0.005, loss=0.189, backward_time=0.234, optim_step_time=0.117, optim0_lr0=7.567e-06, train_time=0.699
|
||||
[bach-gpu011024008134] 2023-04-23 19:00:58,463 (trainer:777) INFO: 2epoch:train:101-150batch:150num_updates: iter_time=1.123e-04, forward_time=0.271, loss_att=0.204, acc=0.942, loss_pre=0.005, loss=0.210, backward_time=0.231, optim_step_time=0.116, optim0_lr0=7.651e-06, train_time=0.692
|
||||
```
|
||||
|
||||
### FAQ
|
||||
### Multi GPUs training and distributed training
|
||||
|
||||
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
|
||||
```
|
||||
@ -1,227 +0,0 @@
|
||||
(简体中文|[English](./quick_start.md))
|
||||
|
||||
# 快速使用
|
||||
|
||||
> **注意**:
|
||||
> modelscope pipeline支持model zoo中的所有模型进行推理和微调。这里我们以typic模型为例来演示用法。
|
||||
|
||||
|
||||
## 使用pipeline进行推理
|
||||
|
||||
### 语音识别
|
||||
#### Paraformer模型
|
||||
```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-pytorch',
|
||||
)
|
||||
|
||||
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)
|
||||
# {'text': '欢迎大家来体验达摩院推出的语音识别模型'}
|
||||
```
|
||||
|
||||
### 语音端点检测
|
||||
#### FSMN-VAD模型
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
from modelscope.utils.logger import get_logger
|
||||
import logging
|
||||
logger = get_logger(log_level=logging.CRITICAL)
|
||||
logger.setLevel(logging.CRITICAL)
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.voice_activity_detection,
|
||||
model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch',
|
||||
)
|
||||
|
||||
segments_result = inference_pipeline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav')
|
||||
print(segments_result)
|
||||
# {'text': [[70, 2340], [2620, 6200], [6480, 23670], [23950, 26250], [26780, 28990], [29950, 31430], [31750, 37600], [38210, 46900], [47310, 49630], [49910, 56460], [56740, 59540], [59820, 70450]]}
|
||||
```
|
||||
|
||||
### 标点恢复
|
||||
#### CT_Transformer模型
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.punctuation,
|
||||
model='damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
|
||||
)
|
||||
|
||||
rec_result = inference_pipeline(text_in='我们都是木头人不会讲话不会动')
|
||||
print(rec_result)
|
||||
# {'text': '我们都是木头人，不会讲话，不会动。'}
|
||||
```
|
||||
|
||||
### 时间戳预测
|
||||
#### TP-Aligner模型
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.speech_timestamp,
|
||||
model='damo/speech_timestamp_prediction-v1-16k-offline',)
|
||||
|
||||
rec_result = inference_pipeline(
|
||||
audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_timestamps.wav',
|
||||
text_in='一 个 东 太 平 洋 国 家 为 什 么 跑 到 西 太 平 洋 来 了 呢',)
|
||||
print(rec_result)
|
||||
# {'text': '<sil> 0.000 0.380;一 0.380 0.560;个 0.560 0.800;东 0.800 0.980;太 0.980 1.140;平 1.140 1.260;洋 1.260 1.440;国 1.440 1.680;家 1.680 1.920;<sil> 1.920 2.040;为 2.040 2.200;什 2.200 2.320;么 2.320 2.500;跑 2.500 2.680;到 2.680 2.860;西 2.860 3.040;太 3.040 3.200;平 3.200 3.380;洋 3.380 3.500;来 3.500 3.640;了 3.640 3.800;呢 3.800 4.150;<sil> 4.150 4.440;', 'timestamp': [[380, 560], [560, 800], [800, 980], [980, 1140], [1140, 1260], [1260, 1440], [1440, 1680], [1680, 1920], [2040, 2200], [2200, 2320], [2320, 2500], [2500, 2680], [2680, 2860], [2860, 3040], [3040, 3200], [3200, 3380], [3380, 3500], [3500, 3640], [3640, 3800], [3800, 4150]]}
|
||||
```
|
||||
|
||||
### 说话人确认
|
||||
#### X-vector模型
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
import numpy as np
|
||||
|
||||
inference_sv_pipline = pipeline(
|
||||
task=Tasks.speaker_verification,
|
||||
model='damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch'
|
||||
)
|
||||
|
||||
# embedding extract
|
||||
spk_embedding = inference_sv_pipline(audio_in='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav')["spk_embedding"]
|
||||
|
||||
# speaker verification
|
||||
rec_result = inference_sv_pipline(audio_in=('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_enroll.wav','https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/sv_example_same.wav'))
|
||||
print(rec_result["scores"][0])
|
||||
# 0.8540499500025098
|
||||
```
|
||||
|
||||
### 说话人日志
|
||||
#### SOND模型
|
||||
```python
|
||||
from modelscope.pipelines import pipeline
|
||||
from modelscope.utils.constant import Tasks
|
||||
|
||||
inference_diar_pipline = pipeline(
|
||||
mode="sond_demo",
|
||||
num_workers=0,
|
||||
task=Tasks.speaker_diarization,
|
||||
diar_model_config="sond.yaml",
|
||||
model='damo/speech_diarization_sond-en-us-callhome-8k-n16k4-pytorch',
|
||||
model_revision="v1.0.3",
|
||||
sv_model="damo/speech_xvector_sv-en-us-callhome-8k-spk6135-pytorch",
|
||||
sv_model_revision="v1.0.0",
|
||||
)
|
||||
|
||||
audio_list=[
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/record.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_A.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_B.wav",
|
||||
"https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_data/spk_B1.wav"
|
||||
]
|
||||
|
||||
results = inference_diar_pipline(audio_in=audio_list)
|
||||
print(results)
|
||||
# {'text': 'spk1 [(0.8, 1.84), (2.8, 6.16), (7.04, 10.64), (12.08, 12.8), (14.24, 15.6)]\nspk2 [(0.0, 1.12), (1.68, 3.2), (4.48, 7.12), (8.48, 9.04), (10.56, 14.48), (15.44, 16.0)]'}
|
||||
```
|
||||
|
||||
### 常见问题
|
||||
#### 使用pipeline进行推理，如何在CPU与GPU进行切换
|
||||
|
||||
The pipeline defaults to decoding with GPU (`ngpu=1`) when GPU is available. If you want to switch to CPU, you could set `ngpu=0`
|
||||
```python
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.auto_speech_recognition,
|
||||
model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch',
|
||||
ngpu=0,
|
||||
)
|
||||
```
|
||||
|
||||
#### 如何从本地模型进行推理（不联网使用）
|
||||
使用modelscope-sdk将模型下载到本地
|
||||
|
||||
```python
|
||||
from modelscope.hub.snapshot_download import snapshot_download
|
||||
|
||||
local_dir_root = "./models_from_modelscope"
|
||||
model_dir = snapshot_download('damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', cache_dir=local_dir_root)
|
||||
```
|
||||
|
||||
或者使用git将模型下载到本地
|
||||
```shell
|
||||
git lfs install
|
||||
# git clone https://www.modelscope.cn/<namespace>/<model-name>.git
|
||||
git clone https://www.modelscope.cn/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch.git
|
||||
```
|
||||
|
||||
从下载的本地模型进行推理（可以不联网使用）
|
||||
```python
|
||||
local_dir_root = "./models_from_modelscope/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
|
||||
inference_pipeline = pipeline(
|
||||
task=Tasks.auto_speech_recognition,
|
||||
model=local_dir_root,
|
||||
)
|
||||
```
|
||||
|
||||
## 使用pipeline进行微调
|
||||
### 语音识别
|
||||
#### Paraformer模型
|
||||
|
||||
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" # 模型保存路径
|
||||
params.data_path = "speech_asr_aishell1_trainsets" # 数据路径，可以为modelscope中已上传数据，也可以是本地数据
|
||||
params.dataset_type = "small" # 小数据量设置small，若数据量大于1000小时，请使用large
|
||||
params.batch_bins = 2000 # batch size，如果dataset_type="small"，batch_bins单位为fbank特征帧数，如果dataset_type="large"，batch_bins单位为毫秒，
|
||||
params.max_epoch = 50 # 最大训练轮数
|
||||
params.lr = 0.00005 # 设置学习率
|
||||
|
||||
modelscope_finetune(params)
|
||||
```
|
||||
|
||||
```shell
|
||||
python finetune.py &> log.txt &
|
||||
```
|
||||
tail log.txt
|
||||
```
|
||||
[bach-gpu011024008134] 2023-04-23 18:59:13,976 (e2e_asr_paraformer:467) INFO: enable sampler in paraformer, sampling_ratio: 0.75
|
||||
[bach-gpu011024008134] 2023-04-23 18:59:48,924 (trainer:777) INFO: 2epoch:train:1-50batch:50num_updates: iter_time=0.008, forward_time=0.302, loss_att=0.186, acc=0.942, loss_pre=0.005, loss=0.192, backward_time=0.231, optim_step_time=0.117, optim0_lr0=7.484e-06, train_time=0.753
|
||||
[bach-gpu011024008134] 2023-04-23 19:00:23,869 (trainer:777) INFO: 2epoch:train:51-100batch:100num_updates: iter_time=1.152e-04, forward_time=0.275, loss_att=0.184, acc=0.945, loss_pre=0.005, loss=0.189, backward_time=0.234, optim_step_time=0.117, optim0_lr0=7.567e-06, train_time=0.699
|
||||
[bach-gpu011024008134] 2023-04-23 19:00:58,463 (trainer:777) INFO: 2epoch:train:101-150batch:150num_updates: iter_time=1.123e-04, forward_time=0.271, loss_att=0.204, acc=0.942, loss_pre=0.005, loss=0.210, backward_time=0.231, optim_step_time=0.116, optim0_lr0=7.651e-06, train_time=0.692
|
||||
```
|
||||
|
||||
### 常见问题
|
||||
### 多GPU训练
|
||||
|
||||
可以使用下面的指令进行多GPU训练
|
||||
```shell
|
||||
CUDA_VISIBLE_DEVICES=1,2 python -m torch.distributed.launch --nproc_per_node 2 finetune.py > log.txt 2>&1
|
||||
```
|
||||
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/speaker_diarization/TEMPLATE/README.md
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/speaker_verification/TEMPLATE/README.md
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/tp/TEMPLATE/README.md
|
||||
@ -1 +0,0 @@
|
||||
../../egs_modelscope/vad/TEMPLATE/README.md
|
||||
Loading…
Reference in New Issue
Block a user