pull/1754/head
TianYuan
3 years ago
commit
a7402203ec
@ -0,0 +1,355 @@
|
|||||||
|
([简体中文](./README_cn.md)|English)
|
||||||
|
|
||||||
|
# Speech Server
|
||||||
|
|
||||||
|
## Introduction
|
||||||
|
This demo is an implementation of starting the streaming speech service and accessing the service. It can be achieved with a single command using `paddlespeech_server` and `paddlespeech_client` or a few lines of code in python.
|
||||||
|
|
||||||
|
|
||||||
|
## Usage
|
||||||
|
### 1. Installation
|
||||||
|
see [installation](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/docs/source/install.md).
|
||||||
|
|
||||||
|
It is recommended to use **paddlepaddle 2.2.1** or above.
|
||||||
|
You can choose one way from meduim and hard to install paddlespeech.
|
||||||
|
|
||||||
|
### 2. Prepare config File
|
||||||
|
The configuration file can be found in `conf/ws_application.yaml` 和 `conf/ws_conformer_application.yaml`.
|
||||||
|
|
||||||
|
At present, the speech tasks integrated by the model include: DeepSpeech2 and conformer.
|
||||||
|
|
||||||
|
|
||||||
|
The input of ASR client demo should be a WAV file(`.wav`), and the sample rate must be the same as the model.
|
||||||
|
|
||||||
|
Here are sample files for thisASR client demo that can be downloaded:
|
||||||
|
```bash
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
|
||||||
|
```
|
||||||
|
|
||||||
|
### 3. Server Usage
|
||||||
|
- Command Line (Recommended)
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# start the service
|
||||||
|
paddlespeech_server start --config_file ./conf/ws_conformer_application.yaml
|
||||||
|
```
|
||||||
|
|
||||||
|
Usage:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
paddlespeech_server start --help
|
||||||
|
```
|
||||||
|
Arguments:
|
||||||
|
- `config_file`: yaml file of the app, defalut: ./conf/ws_conformer_application.yaml
|
||||||
|
- `log_file`: log file. Default: ./log/paddlespeech.log
|
||||||
|
|
||||||
|
Output:
|
||||||
|
```bash
|
||||||
|
[2022-04-21 15:52:18,126] [ INFO] - create the online asr engine instance
|
||||||
|
[2022-04-21 15:52:18,127] [ INFO] - paddlespeech_server set the device: cpu
|
||||||
|
[2022-04-21 15:52:18,128] [ INFO] - Load the pretrained model, tag = conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,128] [ INFO] - File /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/asr1_chunk_conformer_multi_cn_ckpt_0.2.3.model.tar.gz md5 checking...
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - Use pretrained model stored in: /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/model.yaml
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/exp/chunk_conformer/checkpoints/multi_cn.pdparams
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/exp/chunk_conformer/checkpoints/multi_cn.pdparams
|
||||||
|
[2022-04-21 15:52:19,446] [ INFO] - start to create the stream conformer asr engine
|
||||||
|
[2022-04-21 15:52:19,473] [ INFO] - model name: conformer_online
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
[2022-04-21 15:52:21,731] [ INFO] - create the transformer like model success
|
||||||
|
[2022-04-21 15:52:21,733] [ INFO] - Initialize ASR server engine successfully.
|
||||||
|
INFO: Started server process [11173]
|
||||||
|
[2022-04-21 15:52:21] [INFO] [server.py:75] Started server process [11173]
|
||||||
|
INFO: Waiting for application startup.
|
||||||
|
[2022-04-21 15:52:21] [INFO] [on.py:45] Waiting for application startup.
|
||||||
|
INFO: Application startup complete.
|
||||||
|
[2022-04-21 15:52:21] [INFO] [on.py:59] Application startup complete.
|
||||||
|
/home/users/xiongxinlei/.conda/envs/paddlespeech/lib/python3.9/asyncio/base_events.py:1460: DeprecationWarning: The loop argument is deprecated since Python 3.8, and scheduled for removal in Python 3.10.
|
||||||
|
infos = await tasks.gather(*fs, loop=self)
|
||||||
|
/home/users/xiongxinlei/.conda/envs/paddlespeech/lib/python3.9/asyncio/base_events.py:1518: DeprecationWarning: The loop argument is deprecated since Python 3.8, and scheduled for removal in Python 3.10.
|
||||||
|
await tasks.sleep(0, loop=self)
|
||||||
|
INFO: Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
|
||||||
|
[2022-04-21 15:52:21] [INFO] [server.py:206] Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
|
||||||
|
```
|
||||||
|
|
||||||
|
- Python API
|
||||||
|
```python
|
||||||
|
from paddlespeech.server.bin.paddlespeech_server import ServerExecutor
|
||||||
|
|
||||||
|
server_executor = ServerExecutor()
|
||||||
|
server_executor(
|
||||||
|
config_file="./conf/ws_conformer_application.yaml",
|
||||||
|
log_file="./log/paddlespeech.log")
|
||||||
|
```
|
||||||
|
|
||||||
|
Output:
|
||||||
|
```bash
|
||||||
|
[2022-04-21 15:52:18,126] [ INFO] - create the online asr engine instance
|
||||||
|
[2022-04-21 15:52:18,127] [ INFO] - paddlespeech_server set the device: cpu
|
||||||
|
[2022-04-21 15:52:18,128] [ INFO] - Load the pretrained model, tag = conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,128] [ INFO] - File /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/asr1_chunk_conformer_multi_cn_ckpt_0.2.3.model.tar.gz md5 checking...
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - Use pretrained model stored in: /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/model.yaml
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/exp/chunk_conformer/checkpoints/multi_cn.pdparams
|
||||||
|
[2022-04-21 15:52:18,727] [ INFO] - /home/users/xiongxinlei/.paddlespeech/models/conformer_online_multicn-zh-16k/exp/chunk_conformer/checkpoints/multi_cn.pdparams
|
||||||
|
[2022-04-21 15:52:19,446] [ INFO] - start to create the stream conformer asr engine
|
||||||
|
[2022-04-21 15:52:19,473] [ INFO] - model name: conformer_online
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
set kaiming_uniform
|
||||||
|
[2022-04-21 15:52:21,731] [ INFO] - create the transformer like model success
|
||||||
|
[2022-04-21 15:52:21,733] [ INFO] - Initialize ASR server engine successfully.
|
||||||
|
INFO: Started server process [11173]
|
||||||
|
[2022-04-21 15:52:21] [INFO] [server.py:75] Started server process [11173]
|
||||||
|
INFO: Waiting for application startup.
|
||||||
|
[2022-04-21 15:52:21] [INFO] [on.py:45] Waiting for application startup.
|
||||||
|
INFO: Application startup complete.
|
||||||
|
[2022-04-21 15:52:21] [INFO] [on.py:59] Application startup complete.
|
||||||
|
/home/users/xiongxinlei/.conda/envs/paddlespeech/lib/python3.9/asyncio/base_events.py:1460: DeprecationWarning: The loop argument is deprecated since Python 3.8, and scheduled for removal in Python 3.10.
|
||||||
|
infos = await tasks.gather(*fs, loop=self)
|
||||||
|
/home/users/xiongxinlei/.conda/envs/paddlespeech/lib/python3.9/asyncio/base_events.py:1518: DeprecationWarning: The loop argument is deprecated since Python 3.8, and scheduled for removal in Python 3.10.
|
||||||
|
await tasks.sleep(0, loop=self)
|
||||||
|
INFO: Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
|
||||||
|
[2022-04-21 15:52:21] [INFO] [server.py:206] Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
|
||||||
|
```
|
||||||
|
|
||||||
|
|
||||||
|
### 4. ASR Client Usage
|
||||||
|
**Note:** The response time will be slightly longer when using the client for the first time
|
||||||
|
- Command Line (Recommended)
|
||||||
|
```
|
||||||
|
paddlespeech_client asr_online --server_ip 127.0.0.1 --port 8090 --input ./zh.wav
|
||||||
|
```
|
||||||
|
|
||||||
|
Usage:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
paddlespeech_client asr_online --help
|
||||||
|
```
|
||||||
|
Arguments:
|
||||||
|
- `server_ip`: server ip. Default: 127.0.0.1
|
||||||
|
- `port`: server port. Default: 8090
|
||||||
|
- `input`(required): Audio file to be recognized.
|
||||||
|
- `sample_rate`: Audio ampling rate, default: 16000.
|
||||||
|
- `lang`: Language. Default: "zh_cn".
|
||||||
|
- `audio_format`: Audio format. Default: "wav".
|
||||||
|
|
||||||
|
Output:
|
||||||
|
```bash
|
||||||
|
[2022-04-21 15:59:03,904] [ INFO] - receive msg={"status": "ok", "signal": "server_ready"}
|
||||||
|
[2022-04-21 15:59:03,960] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:03,973] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:03,987] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,000] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,012] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,024] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,036] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,047] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,607] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,620] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,633] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,645] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,657] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,669] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,680] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:05,176] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,185] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,192] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,200] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,208] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,216] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,224] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,232] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,724] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,732] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,740] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,747] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,755] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,763] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,770] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:06,271] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,279] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,287] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,294] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,302] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,310] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,318] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,326] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,833] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,842] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,850] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,858] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,866] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,874] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,882] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:07,400] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,408] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,416] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,424] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,432] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,440] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,447] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,455] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,984] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:07,992] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,001] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,008] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,016] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,024] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:12,883] [ INFO] - final receive msg={'status': 'ok', 'signal': 'finished', 'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:12,884] [ INFO] - 我认为跑步最重要的就是给我带来了身体健康
|
||||||
|
[2022-04-21 15:59:12,884] [ INFO] - Response time 9.051567 s.
|
||||||
|
|
||||||
|
```
|
||||||
|
|
||||||
|
- Python API
|
||||||
|
```python
|
||||||
|
from paddlespeech.server.bin.paddlespeech_client import ASRClientExecutor
|
||||||
|
import json
|
||||||
|
|
||||||
|
asrclient_executor = ASRClientExecutor()
|
||||||
|
res = asrclient_executor(
|
||||||
|
input="./zh.wav",
|
||||||
|
server_ip="127.0.0.1",
|
||||||
|
port=8090,
|
||||||
|
sample_rate=16000,
|
||||||
|
lang="zh_cn",
|
||||||
|
audio_format="wav")
|
||||||
|
print(res.json())
|
||||||
|
```
|
||||||
|
|
||||||
|
Output:
|
||||||
|
```bash
|
||||||
|
[2022-04-21 15:59:03,904] [ INFO] - receive msg={"status": "ok", "signal": "server_ready"}
|
||||||
|
[2022-04-21 15:59:03,960] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:03,973] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:03,987] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,000] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,012] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,024] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,036] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,047] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,607] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,620] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,633] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,645] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,657] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,669] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:04,680] [ INFO] - receive msg={'asr_results': ''}
|
||||||
|
[2022-04-21 15:59:05,176] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,185] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,192] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,200] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,208] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,216] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,224] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,232] [ INFO] - receive msg={'asr_results': '我认为跑'}
|
||||||
|
[2022-04-21 15:59:05,724] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,732] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,740] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,747] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,755] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,763] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:05,770] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的'}
|
||||||
|
[2022-04-21 15:59:06,271] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,279] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,287] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,294] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,302] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,310] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,318] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,326] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是'}
|
||||||
|
[2022-04-21 15:59:06,833] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,842] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,850] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,858] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,866] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,874] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:06,882] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给'}
|
||||||
|
[2022-04-21 15:59:07,400] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,408] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,416] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,424] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,432] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,440] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,447] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,455] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了'}
|
||||||
|
[2022-04-21 15:59:07,984] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:07,992] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,001] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,008] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,016] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:08,024] [ INFO] - receive msg={'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:12,883] [ INFO] - final receive msg={'status': 'ok', 'signal': 'finished', 'asr_results': '我认为跑步最重要的就是给我带来了身体健康'}
|
||||||
|
[2022-04-21 15:59:12,884] [ INFO] - 我认为跑步最重要的就是给我带来了身体健康
|
||||||
|
```
|
||||||
@ -0,0 +1,47 @@
|
|||||||
|
# This is the parameter configuration file for PaddleSpeech Serving.
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# SERVER SETTING #
|
||||||
|
#################################################################################
|
||||||
|
host: 0.0.0.0
|
||||||
|
port: 8090
|
||||||
|
|
||||||
|
# The task format in the engin_list is: <speech task>_<engine type>
|
||||||
|
# task choices = ['asr_online', 'tts_online']
|
||||||
|
# protocol = ['websocket', 'http'] (only one can be selected).
|
||||||
|
# websocket only support online engine type.
|
||||||
|
protocol: 'websocket'
|
||||||
|
engine_list: ['asr_online']
|
||||||
|
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# ENGINE CONFIG #
|
||||||
|
#################################################################################
|
||||||
|
|
||||||
|
################################### ASR #########################################
|
||||||
|
################### speech task: asr; engine_type: online #######################
|
||||||
|
asr_online:
|
||||||
|
model_type: 'deepspeech2online_aishell'
|
||||||
|
am_model: # the pdmodel file of am static model [optional]
|
||||||
|
am_params: # the pdiparams file of am static model [optional]
|
||||||
|
lang: 'zh'
|
||||||
|
sample_rate: 16000
|
||||||
|
cfg_path:
|
||||||
|
decode_method:
|
||||||
|
force_yes: True
|
||||||
|
|
||||||
|
am_predictor_conf:
|
||||||
|
device: # set 'gpu:id' or 'cpu'
|
||||||
|
switch_ir_optim: True
|
||||||
|
glog_info: False # True -> print glog
|
||||||
|
summary: True # False -> do not show predictor config
|
||||||
|
|
||||||
|
chunk_buffer_conf:
|
||||||
|
frame_duration_ms: 80
|
||||||
|
shift_ms: 40
|
||||||
|
sample_rate: 16000
|
||||||
|
sample_width: 2
|
||||||
|
window_n: 7 # frame
|
||||||
|
shift_n: 4 # frame
|
||||||
|
window_ms: 20 # ms
|
||||||
|
shift_ms: 10 # ms
|
||||||
@ -0,0 +1,45 @@
|
|||||||
|
# This is the parameter configuration file for PaddleSpeech Serving.
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# SERVER SETTING #
|
||||||
|
#################################################################################
|
||||||
|
host: 0.0.0.0
|
||||||
|
port: 8090
|
||||||
|
|
||||||
|
# The task format in the engin_list is: <speech task>_<engine type>
|
||||||
|
# task choices = ['asr_online', 'tts_online']
|
||||||
|
# protocol = ['websocket', 'http'] (only one can be selected).
|
||||||
|
# websocket only support online engine type.
|
||||||
|
protocol: 'websocket'
|
||||||
|
engine_list: ['asr_online']
|
||||||
|
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# ENGINE CONFIG #
|
||||||
|
#################################################################################
|
||||||
|
|
||||||
|
################################### ASR #########################################
|
||||||
|
################### speech task: asr; engine_type: online #######################
|
||||||
|
asr_online:
|
||||||
|
model_type: 'conformer_online_multicn'
|
||||||
|
am_model: # the pdmodel file of am static model [optional]
|
||||||
|
am_params: # the pdiparams file of am static model [optional]
|
||||||
|
lang: 'zh'
|
||||||
|
sample_rate: 16000
|
||||||
|
cfg_path:
|
||||||
|
decode_method:
|
||||||
|
force_yes: True
|
||||||
|
device: # cpu or gpu:id
|
||||||
|
am_predictor_conf:
|
||||||
|
device: # set 'gpu:id' or 'cpu'
|
||||||
|
switch_ir_optim: True
|
||||||
|
glog_info: False # True -> print glog
|
||||||
|
summary: True # False -> do not show predictor config
|
||||||
|
|
||||||
|
chunk_buffer_conf:
|
||||||
|
window_n: 7 # frame
|
||||||
|
shift_n: 4 # frame
|
||||||
|
window_ms: 25 # ms
|
||||||
|
shift_ms: 10 # ms
|
||||||
|
sample_rate: 16000
|
||||||
|
sample_width: 2
|
||||||
@ -0,0 +1,2 @@
|
|||||||
|
# start the streaming asr service
|
||||||
|
paddlespeech_server start --config_file ./conf/ws_conformer_application.yaml
|
||||||
@ -0,0 +1,5 @@
|
|||||||
|
# download the test wav
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
|
||||||
|
|
||||||
|
# read the wav and pass it to service
|
||||||
|
python3 websocket_client.py --wavfile ./zh.wav
|
||||||
|
Before Width: | Height: | Size: 949 KiB After Width: | Height: | Size: 949 KiB |
|
Before Width: | Height: | Size: 432 KiB After Width: | Height: | Size: 432 KiB |
|
Before Width: | Height: | Size: 72 KiB After Width: | Height: | Size: 72 KiB |
|
Before Width: | Height: | Size: 286 KiB After Width: | Height: | Size: 286 KiB |
|
Before Width: | Height: | Size: 4.2 KiB After Width: | Height: | Size: 4.2 KiB |
@ -0,0 +1,62 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
#!/usr/bin/python
|
||||||
|
# -*- coding: UTF-8 -*-
|
||||||
|
import argparse
|
||||||
|
import asyncio
|
||||||
|
import codecs
|
||||||
|
import logging
|
||||||
|
import os
|
||||||
|
|
||||||
|
from paddlespeech.cli.log import logger
|
||||||
|
from paddlespeech.server.utils.audio_handler import ASRAudioHandler
|
||||||
|
|
||||||
|
|
||||||
|
def main(args):
|
||||||
|
logger.info("asr websocket client start")
|
||||||
|
handler = ASRAudioHandler("127.0.0.1", 8090)
|
||||||
|
loop = asyncio.get_event_loop()
|
||||||
|
|
||||||
|
# support to process single audio file
|
||||||
|
if args.wavfile and os.path.exists(args.wavfile):
|
||||||
|
logger.info(f"start to process the wavscp: {args.wavfile}")
|
||||||
|
result = loop.run_until_complete(handler.run(args.wavfile))
|
||||||
|
result = result["asr_results"]
|
||||||
|
logger.info(f"asr websocket client finished : {result}")
|
||||||
|
|
||||||
|
# support to process batch audios from wav.scp
|
||||||
|
if args.wavscp and os.path.exists(args.wavscp):
|
||||||
|
logging.info(f"start to process the wavscp: {args.wavscp}")
|
||||||
|
with codecs.open(args.wavscp, 'r', encoding='utf-8') as f,\
|
||||||
|
codecs.open("result.txt", 'w', encoding='utf-8') as w:
|
||||||
|
for line in f:
|
||||||
|
utt_name, utt_path = line.strip().split()
|
||||||
|
result = loop.run_until_complete(handler.run(utt_path))
|
||||||
|
result = result["asr_results"]
|
||||||
|
w.write(f"{utt_name} {result}\n")
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
logger.info("Start to do streaming asr client")
|
||||||
|
parser = argparse.ArgumentParser()
|
||||||
|
parser.add_argument(
|
||||||
|
"--wavfile",
|
||||||
|
action="store",
|
||||||
|
help="wav file path ",
|
||||||
|
default="./16_audio.wav")
|
||||||
|
parser.add_argument(
|
||||||
|
"--wavscp", type=str, default=None, help="The batch audios dict text")
|
||||||
|
args = parser.parse_args()
|
||||||
|
|
||||||
|
main(args)
|
||||||
@ -0,0 +1,45 @@
|
|||||||
|
# This is the parameter configuration file for PaddleSpeech Serving.
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# SERVER SETTING #
|
||||||
|
#################################################################################
|
||||||
|
host: 0.0.0.0
|
||||||
|
port: 8090
|
||||||
|
|
||||||
|
# The task format in the engin_list is: <speech task>_<engine type>
|
||||||
|
# task choices = ['asr_online', 'tts_online']
|
||||||
|
# protocol = ['websocket', 'http'] (only one can be selected).
|
||||||
|
# websocket only support online engine type.
|
||||||
|
protocol: 'websocket'
|
||||||
|
engine_list: ['asr_online']
|
||||||
|
|
||||||
|
|
||||||
|
#################################################################################
|
||||||
|
# ENGINE CONFIG #
|
||||||
|
#################################################################################
|
||||||
|
|
||||||
|
################################### ASR #########################################
|
||||||
|
################### speech task: asr; engine_type: online #######################
|
||||||
|
asr_online:
|
||||||
|
model_type: 'conformer_online_multicn'
|
||||||
|
am_model: # the pdmodel file of am static model [optional]
|
||||||
|
am_params: # the pdiparams file of am static model [optional]
|
||||||
|
lang: 'zh'
|
||||||
|
sample_rate: 16000
|
||||||
|
cfg_path:
|
||||||
|
decode_method:
|
||||||
|
force_yes: True
|
||||||
|
device: # cpu or gpu:id
|
||||||
|
am_predictor_conf:
|
||||||
|
device: # set 'gpu:id' or 'cpu'
|
||||||
|
switch_ir_optim: True
|
||||||
|
glog_info: False # True -> print glog
|
||||||
|
summary: True # False -> do not show predictor config
|
||||||
|
|
||||||
|
chunk_buffer_conf:
|
||||||
|
window_n: 7 # frame
|
||||||
|
shift_n: 4 # frame
|
||||||
|
window_ms: 25 # ms
|
||||||
|
shift_ms: 10 # ms
|
||||||
|
sample_rate: 16000
|
||||||
|
sample_width: 2
|
||||||
@ -0,0 +1,130 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
from collections import defaultdict
|
||||||
|
|
||||||
|
import paddle
|
||||||
|
|
||||||
|
from paddlespeech.cli.log import logger
|
||||||
|
from paddlespeech.s2t.utils.utility import log_add
|
||||||
|
|
||||||
|
__all__ = ['CTCPrefixBeamSearch']
|
||||||
|
|
||||||
|
|
||||||
|
class CTCPrefixBeamSearch:
|
||||||
|
def __init__(self, config):
|
||||||
|
"""Implement the ctc prefix beam search
|
||||||
|
|
||||||
|
Args:
|
||||||
|
config (yacs.config.CfgNode): _description_
|
||||||
|
"""
|
||||||
|
self.config = config
|
||||||
|
self.reset()
|
||||||
|
|
||||||
|
@paddle.no_grad()
|
||||||
|
def search(self, ctc_probs, device, blank_id=0):
|
||||||
|
"""ctc prefix beam search method decode a chunk feature
|
||||||
|
|
||||||
|
Args:
|
||||||
|
xs (paddle.Tensor): feature data
|
||||||
|
ctc_probs (paddle.Tensor): the ctc probability of all the tokens
|
||||||
|
device (paddle.fluid.core_avx.Place): the feature host device, such as CUDAPlace(0).
|
||||||
|
blank_id (int, optional): the blank id in the vocab. Defaults to 0.
|
||||||
|
|
||||||
|
Returns:
|
||||||
|
list: the search result
|
||||||
|
"""
|
||||||
|
# decode
|
||||||
|
logger.info("start to ctc prefix search")
|
||||||
|
|
||||||
|
batch_size = 1
|
||||||
|
beam_size = self.config.beam_size
|
||||||
|
maxlen = ctc_probs.shape[0]
|
||||||
|
|
||||||
|
assert len(ctc_probs.shape) == 2
|
||||||
|
|
||||||
|
# cur_hyps: (prefix, (blank_ending_score, none_blank_ending_score))
|
||||||
|
# blank_ending_score and none_blank_ending_score in ln domain
|
||||||
|
if self.cur_hyps is None:
|
||||||
|
self.cur_hyps = [(tuple(), (0.0, -float('inf')))]
|
||||||
|
# 2. CTC beam search step by step
|
||||||
|
for t in range(0, maxlen):
|
||||||
|
logp = ctc_probs[t] # (vocab_size,)
|
||||||
|
# key: prefix, value (pb, pnb), default value(-inf, -inf)
|
||||||
|
next_hyps = defaultdict(lambda: (-float('inf'), -float('inf')))
|
||||||
|
|
||||||
|
# 2.1 First beam prune: select topk best
|
||||||
|
# do token passing process
|
||||||
|
top_k_logp, top_k_index = logp.topk(beam_size) # (beam_size,)
|
||||||
|
for s in top_k_index:
|
||||||
|
s = s.item()
|
||||||
|
ps = logp[s].item()
|
||||||
|
for prefix, (pb, pnb) in self.cur_hyps:
|
||||||
|
last = prefix[-1] if len(prefix) > 0 else None
|
||||||
|
if s == blank_id: # blank
|
||||||
|
n_pb, n_pnb = next_hyps[prefix]
|
||||||
|
n_pb = log_add([n_pb, pb + ps, pnb + ps])
|
||||||
|
next_hyps[prefix] = (n_pb, n_pnb)
|
||||||
|
elif s == last:
|
||||||
|
# Update *ss -> *s;
|
||||||
|
n_pb, n_pnb = next_hyps[prefix]
|
||||||
|
n_pnb = log_add([n_pnb, pnb + ps])
|
||||||
|
next_hyps[prefix] = (n_pb, n_pnb)
|
||||||
|
# Update *s-s -> *ss, - is for blank
|
||||||
|
n_prefix = prefix + (s, )
|
||||||
|
n_pb, n_pnb = next_hyps[n_prefix]
|
||||||
|
n_pnb = log_add([n_pnb, pb + ps])
|
||||||
|
next_hyps[n_prefix] = (n_pb, n_pnb)
|
||||||
|
else:
|
||||||
|
n_prefix = prefix + (s, )
|
||||||
|
n_pb, n_pnb = next_hyps[n_prefix]
|
||||||
|
n_pnb = log_add([n_pnb, pb + ps, pnb + ps])
|
||||||
|
next_hyps[n_prefix] = (n_pb, n_pnb)
|
||||||
|
|
||||||
|
# 2.2 Second beam prune
|
||||||
|
next_hyps = sorted(
|
||||||
|
next_hyps.items(),
|
||||||
|
key=lambda x: log_add(list(x[1])),
|
||||||
|
reverse=True)
|
||||||
|
self.cur_hyps = next_hyps[:beam_size]
|
||||||
|
|
||||||
|
self.hyps = [(y[0], log_add([y[1][0], y[1][1]])) for y in self.cur_hyps]
|
||||||
|
logger.info("ctc prefix search success")
|
||||||
|
return self.hyps
|
||||||
|
|
||||||
|
def get_one_best_hyps(self):
|
||||||
|
"""Return the one best result
|
||||||
|
|
||||||
|
Returns:
|
||||||
|
list: the one best result
|
||||||
|
"""
|
||||||
|
return [self.hyps[0][0]]
|
||||||
|
|
||||||
|
def get_hyps(self):
|
||||||
|
"""Return the search hyps
|
||||||
|
|
||||||
|
Returns:
|
||||||
|
list: return the search hyps
|
||||||
|
"""
|
||||||
|
return self.hyps
|
||||||
|
|
||||||
|
def reset(self):
|
||||||
|
"""Rest the search cache value
|
||||||
|
"""
|
||||||
|
self.cur_hyps = None
|
||||||
|
self.hyps = None
|
||||||
|
|
||||||
|
def finalize_search(self):
|
||||||
|
"""do nothing in ctc_prefix_beam_search
|
||||||
|
"""
|
||||||
|
pass
|
||||||
@ -0,0 +1,13 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
@ -0,0 +1,13 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
@ -0,0 +1,13 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
@ -0,0 +1,35 @@
|
|||||||
|
([简体中文](./README_cn.md)|English)
|
||||||
|
|
||||||
|
# Speech Service
|
||||||
|
|
||||||
|
## Introduction
|
||||||
|
|
||||||
|
This document introduces a client for streaming asr service: microphone
|
||||||
|
|
||||||
|
|
||||||
|
## Usage
|
||||||
|
### 1. Install
|
||||||
|
Refer [Install](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/docs/source/install.md).
|
||||||
|
|
||||||
|
**paddlepaddle 2.2.1** 或以上版本。
|
||||||
|
It is recommended to use **paddlepaddle 2.2.1** or above.
|
||||||
|
You can choose one way from meduim and hard to install paddlespeech.
|
||||||
|
|
||||||
|
|
||||||
|
### 2. Prepare config File
|
||||||
|
|
||||||
|
|
||||||
|
The input of ASR client demo should be a WAV file(`.wav`), and the sample rate must be the same as the model.
|
||||||
|
|
||||||
|
Here are sample files for thisASR client demo that can be downloaded:
|
||||||
|
```bash
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
|
||||||
|
```
|
||||||
|
|
||||||
|
### 2. Streaming ASR Client Usage
|
||||||
|
|
||||||
|
- microphone
|
||||||
|
```
|
||||||
|
python microphone_client.py
|
||||||
|
|
||||||
|
```
|
||||||
@ -1,136 +0,0 @@
|
|||||||
# Copyright (c) 2022 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.
|
|
||||||
# You may obtain a copy of the License at
|
|
||||||
#
|
|
||||||
# http://www.apache.org/licenses/LICENSE-2.0
|
|
||||||
#
|
|
||||||
# Unless required by applicable law or agreed to in writing, software
|
|
||||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
||||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
||||||
# See the License for the specific language governing permissions and
|
|
||||||
# limitations under the License.
|
|
||||||
#!/usr/bin/python
|
|
||||||
# -*- coding: UTF-8 -*-
|
|
||||||
import argparse
|
|
||||||
import asyncio
|
|
||||||
import codecs
|
|
||||||
import json
|
|
||||||
import logging
|
|
||||||
import os
|
|
||||||
|
|
||||||
import numpy as np
|
|
||||||
import soundfile
|
|
||||||
import websockets
|
|
||||||
|
|
||||||
|
|
||||||
class ASRAudioHandler:
|
|
||||||
def __init__(self, url="127.0.0.1", port=8090):
|
|
||||||
self.url = url
|
|
||||||
self.port = port
|
|
||||||
self.url = "ws://" + self.url + ":" + str(self.port) + "/ws/asr"
|
|
||||||
|
|
||||||
def read_wave(self, wavfile_path: str):
|
|
||||||
samples, sample_rate = soundfile.read(wavfile_path, dtype='int16')
|
|
||||||
x_len = len(samples)
|
|
||||||
# chunk_stride = 40 * 16 #40ms, sample_rate = 16kHz
|
|
||||||
chunk_size = 80 * 16 #80ms, sample_rate = 16kHz
|
|
||||||
|
|
||||||
if x_len % chunk_size != 0:
|
|
||||||
padding_len_x = chunk_size - x_len % chunk_size
|
|
||||||
else:
|
|
||||||
padding_len_x = 0
|
|
||||||
|
|
||||||
padding = np.zeros((padding_len_x), dtype=samples.dtype)
|
|
||||||
padded_x = np.concatenate([samples, padding], axis=0)
|
|
||||||
|
|
||||||
assert (x_len + padding_len_x) % chunk_size == 0
|
|
||||||
num_chunk = (x_len + padding_len_x) / chunk_size
|
|
||||||
num_chunk = int(num_chunk)
|
|
||||||
|
|
||||||
for i in range(0, num_chunk):
|
|
||||||
start = i * chunk_size
|
|
||||||
end = start + chunk_size
|
|
||||||
x_chunk = padded_x[start:end]
|
|
||||||
yield x_chunk
|
|
||||||
|
|
||||||
async def run(self, wavfile_path: str):
|
|
||||||
logging.info("send a message to the server")
|
|
||||||
async with websockets.connect(self.url) as ws:
|
|
||||||
audio_info = json.dumps(
|
|
||||||
{
|
|
||||||
"name": "test.wav",
|
|
||||||
"signal": "start",
|
|
||||||
"nbest": 5
|
|
||||||
},
|
|
||||||
sort_keys=True,
|
|
||||||
indent=4,
|
|
||||||
separators=(',', ': '))
|
|
||||||
await ws.send(audio_info)
|
|
||||||
msg = await ws.recv()
|
|
||||||
logging.info("receive msg={}".format(msg))
|
|
||||||
|
|
||||||
# send chunk audio data to engine
|
|
||||||
for chunk_data in self.read_wave(wavfile_path):
|
|
||||||
await ws.send(chunk_data.tobytes())
|
|
||||||
msg = await ws.recv()
|
|
||||||
msg = json.loads(msg)
|
|
||||||
logging.info("receive msg={}".format(msg))
|
|
||||||
|
|
||||||
result = msg
|
|
||||||
# finished
|
|
||||||
audio_info = json.dumps(
|
|
||||||
{
|
|
||||||
"name": "test.wav",
|
|
||||||
"signal": "end",
|
|
||||||
"nbest": 5
|
|
||||||
},
|
|
||||||
sort_keys=True,
|
|
||||||
indent=4,
|
|
||||||
separators=(',', ': '))
|
|
||||||
await ws.send(audio_info)
|
|
||||||
msg = await ws.recv()
|
|
||||||
msg = json.loads(msg)
|
|
||||||
logging.info("receive msg={}".format(msg))
|
|
||||||
|
|
||||||
return result
|
|
||||||
|
|
||||||
|
|
||||||
def main(args):
|
|
||||||
logging.basicConfig(level=logging.INFO)
|
|
||||||
logging.info("asr websocket client start")
|
|
||||||
handler = ASRAudioHandler("127.0.0.1", 8090)
|
|
||||||
loop = asyncio.get_event_loop()
|
|
||||||
|
|
||||||
# support to process single audio file
|
|
||||||
if args.wavfile and os.path.exists(args.wavfile):
|
|
||||||
logging.info(f"start to process the wavscp: {args.wavfile}")
|
|
||||||
result = loop.run_until_complete(handler.run(args.wavfile))
|
|
||||||
result = result["asr_results"]
|
|
||||||
logging.info(f"asr websocket client finished : {result}")
|
|
||||||
|
|
||||||
# support to process batch audios from wav.scp
|
|
||||||
if args.wavscp and os.path.exists(args.wavscp):
|
|
||||||
logging.info(f"start to process the wavscp: {args.wavscp}")
|
|
||||||
with codecs.open(args.wavscp, 'r', encoding='utf-8') as f,\
|
|
||||||
codecs.open("result.txt", 'w', encoding='utf-8') as w:
|
|
||||||
for line in f:
|
|
||||||
utt_name, utt_path = line.strip().split()
|
|
||||||
result = loop.run_until_complete(handler.run(utt_path))
|
|
||||||
result = result["asr_results"]
|
|
||||||
w.write(f"{utt_name} {result}\n")
|
|
||||||
|
|
||||||
|
|
||||||
if __name__ == "__main__":
|
|
||||||
parser = argparse.ArgumentParser()
|
|
||||||
parser.add_argument(
|
|
||||||
"--wavfile",
|
|
||||||
action="store",
|
|
||||||
help="wav file path ",
|
|
||||||
default="./16_audio.wav")
|
|
||||||
parser.add_argument(
|
|
||||||
"--wavscp", type=str, default=None, help="The batch audios dict text")
|
|
||||||
args = parser.parse_args()
|
|
||||||
|
|
||||||
main(args)
|
|
||||||
@ -0,0 +1,76 @@
|
|||||||
|
# Copyright (c) 2022 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.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
import paddle
|
||||||
|
import paddle.nn as nn
|
||||||
|
from paddle.autograd import PyLayer
|
||||||
|
|
||||||
|
|
||||||
|
class GradientReversalFunction(PyLayer):
|
||||||
|
"""Gradient Reversal Layer from:
|
||||||
|
Unsupervised Domain Adaptation by Backpropagation (Ganin & Lempitsky, 2015)
|
||||||
|
|
||||||
|
Forward pass is the identity function. In the backward pass,
|
||||||
|
the upstream gradients are multiplied by -lambda (i.e. gradient is reversed)
|
||||||
|
"""
|
||||||
|
|
||||||
|
@staticmethod
|
||||||
|
def forward(ctx, x, lambda_=1):
|
||||||
|
"""Forward in networks
|
||||||
|
"""
|
||||||
|
ctx.save_for_backward(lambda_)
|
||||||
|
return x.clone()
|
||||||
|
|
||||||
|
@staticmethod
|
||||||
|
def backward(ctx, grads):
|
||||||
|
"""Backward in networks
|
||||||
|
"""
|
||||||
|
lambda_, = ctx.saved_tensor()
|
||||||
|
dx = -lambda_ * grads
|
||||||
|
return dx
|
||||||
|
|
||||||
|
|
||||||
|
class GradientReversalLayer(nn.Layer):
|
||||||
|
"""Gradient Reversal Layer from:
|
||||||
|
Unsupervised Domain Adaptation by Backpropagation (Ganin & Lempitsky, 2015)
|
||||||
|
|
||||||
|
Forward pass is the identity function. In the backward pass,
|
||||||
|
the upstream gradients are multiplied by -lambda (i.e. gradient is reversed)
|
||||||
|
"""
|
||||||
|
|
||||||
|
def __init__(self, lambda_=1):
|
||||||
|
super(GradientReversalLayer, self).__init__()
|
||||||
|
self.lambda_ = lambda_
|
||||||
|
|
||||||
|
def forward(self, x):
|
||||||
|
"""Forward in networks
|
||||||
|
"""
|
||||||
|
return GradientReversalFunction.apply(x, self.lambda_)
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
paddle.set_device("cpu")
|
||||||
|
|
||||||
|
data = paddle.randn([2, 3], dtype="float64")
|
||||||
|
data.stop_gradient = False
|
||||||
|
grl = GradientReversalLayer(1)
|
||||||
|
out = grl(data)
|
||||||
|
out.mean().backward()
|
||||||
|
print(data.grad)
|
||||||
|
|
||||||
|
data = paddle.randn([2, 3], dtype="float64")
|
||||||
|
data.stop_gradient = False
|
||||||
|
grl = GradientReversalLayer(-1)
|
||||||
|
out = grl(data)
|
||||||
|
out.mean().backward()
|
||||||
|
print(data.grad)
|
||||||
@ -0,0 +1,88 @@
|
|||||||
|
// Copyright (c) 2022 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.
|
||||||
|
// You may obtain a copy of the License at
|
||||||
|
//
|
||||||
|
// http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
//
|
||||||
|
// Unless required by applicable law or agreed to in writing, software
|
||||||
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
// See the License for the specific language governing permissions and
|
||||||
|
// limitations under the License.
|
||||||
|
|
||||||
|
#include "decoder/recognizer.h"
|
||||||
|
#include "decoder/param.h"
|
||||||
|
#include "kaldi/feat/wave-reader.h"
|
||||||
|
#include "kaldi/util/table-types.h"
|
||||||
|
|
||||||
|
DEFINE_string(wav_rspecifier, "", "test feature rspecifier");
|
||||||
|
DEFINE_string(result_wspecifier, "", "test result wspecifier");
|
||||||
|
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
gflags::ParseCommandLineFlags(&argc, &argv, false);
|
||||||
|
google::InitGoogleLogging(argv[0]);
|
||||||
|
|
||||||
|
ppspeech::RecognizerResource resource = ppspeech::InitRecognizerResoure();
|
||||||
|
ppspeech::Recognizer recognizer(resource);
|
||||||
|
|
||||||
|
kaldi::SequentialTableReader<kaldi::WaveHolder> wav_reader(
|
||||||
|
FLAGS_wav_rspecifier);
|
||||||
|
kaldi::TokenWriter result_writer(FLAGS_result_wspecifier);
|
||||||
|
int sample_rate = 16000;
|
||||||
|
float streaming_chunk = FLAGS_streaming_chunk;
|
||||||
|
int chunk_sample_size = streaming_chunk * sample_rate;
|
||||||
|
LOG(INFO) << "sr: " << sample_rate;
|
||||||
|
LOG(INFO) << "chunk size (s): " << streaming_chunk;
|
||||||
|
LOG(INFO) << "chunk size (sample): " << chunk_sample_size;
|
||||||
|
|
||||||
|
int32 num_done = 0, num_err = 0;
|
||||||
|
|
||||||
|
for (; !wav_reader.Done(); wav_reader.Next()) {
|
||||||
|
std::string utt = wav_reader.Key();
|
||||||
|
const kaldi::WaveData& wave_data = wav_reader.Value();
|
||||||
|
|
||||||
|
int32 this_channel = 0;
|
||||||
|
kaldi::SubVector<kaldi::BaseFloat> waveform(wave_data.Data(),
|
||||||
|
this_channel);
|
||||||
|
int tot_samples = waveform.Dim();
|
||||||
|
LOG(INFO) << "wav len (sample): " << tot_samples;
|
||||||
|
|
||||||
|
int sample_offset = 0;
|
||||||
|
std::vector<kaldi::Vector<BaseFloat>> feats;
|
||||||
|
int feature_rows = 0;
|
||||||
|
while (sample_offset < tot_samples) {
|
||||||
|
int cur_chunk_size =
|
||||||
|
std::min(chunk_sample_size, tot_samples - sample_offset);
|
||||||
|
|
||||||
|
kaldi::Vector<kaldi::BaseFloat> wav_chunk(cur_chunk_size);
|
||||||
|
for (int i = 0; i < cur_chunk_size; ++i) {
|
||||||
|
wav_chunk(i) = waveform(sample_offset + i);
|
||||||
|
}
|
||||||
|
// wav_chunk = waveform.Range(sample_offset + i, cur_chunk_size);
|
||||||
|
|
||||||
|
recognizer.Accept(wav_chunk);
|
||||||
|
if (cur_chunk_size < chunk_sample_size) {
|
||||||
|
recognizer.SetFinished();
|
||||||
|
}
|
||||||
|
recognizer.Decode();
|
||||||
|
|
||||||
|
// no overlap
|
||||||
|
sample_offset += cur_chunk_size;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string result;
|
||||||
|
result = recognizer.GetFinalResult();
|
||||||
|
recognizer.Reset();
|
||||||
|
if (result.empty()) {
|
||||||
|
// the TokenWriter can not write empty string.
|
||||||
|
++num_err;
|
||||||
|
KALDI_LOG << " the result of " << utt << " is empty";
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
KALDI_LOG << " the result of " << utt << " is " << result;
|
||||||
|
result_writer.Write(utt, result);
|
||||||
|
++num_done;
|
||||||
|
}
|
||||||
|
}
|
||||||
@ -0,0 +1,2 @@
|
|||||||
|
data
|
||||||
|
exp
|
||||||
@ -0,0 +1,9 @@
|
|||||||
|
cmake_minimum_required(VERSION 3.14 FATAL_ERROR)
|
||||||
|
|
||||||
|
add_executable(websocket_server_main ${CMAKE_CURRENT_SOURCE_DIR}/websocket_server_main.cc)
|
||||||
|
target_include_directories(websocket_server_main PRIVATE ${SPEECHX_ROOT} ${SPEECHX_ROOT}/kaldi)
|
||||||
|
target_link_libraries(websocket_server_main PUBLIC frontend kaldi-feat-common nnet decoder fst utils gflags glog kaldi-base kaldi-matrix kaldi-util kaldi-decoder websocket ${DEPS})
|
||||||
|
|
||||||
|
add_executable(websocket_client_main ${CMAKE_CURRENT_SOURCE_DIR}/websocket_client_main.cc)
|
||||||
|
target_include_directories(websocket_client_main PRIVATE ${SPEECHX_ROOT} ${SPEECHX_ROOT}/kaldi)
|
||||||
|
target_link_libraries(websocket_client_main PUBLIC frontend kaldi-feat-common nnet decoder fst utils gflags glog kaldi-base kaldi-matrix kaldi-util kaldi-decoder websocket ${DEPS})
|
||||||
@ -0,0 +1,14 @@
|
|||||||
|
# This contains the locations of binarys build required for running the examples.
|
||||||
|
|
||||||
|
SPEECHX_ROOT=$PWD/../../..
|
||||||
|
SPEECHX_EXAMPLES=$SPEECHX_ROOT/build/examples
|
||||||
|
|
||||||
|
SPEECHX_TOOLS=$SPEECHX_ROOT/tools
|
||||||
|
TOOLS_BIN=$SPEECHX_TOOLS/valgrind/install/bin
|
||||||
|
|
||||||
|
[ -d $SPEECHX_EXAMPLES ] || { echo "Error: 'build/examples' directory not found. please ensure that the project build successfully"; }
|
||||||
|
|
||||||
|
export LC_AL=C
|
||||||
|
|
||||||
|
SPEECHX_BIN=$SPEECHX_EXAMPLES/ds2_ol/websocket:$SPEECHX_EXAMPLES/ds2_ol/feat
|
||||||
|
export PATH=$PATH:$SPEECHX_BIN:$TOOLS_BIN
|
||||||
@ -0,0 +1,37 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
set +x
|
||||||
|
set -e
|
||||||
|
|
||||||
|
. path.sh
|
||||||
|
|
||||||
|
# 1. compile
|
||||||
|
if [ ! -d ${SPEECHX_EXAMPLES} ]; then
|
||||||
|
pushd ${SPEECHX_ROOT}
|
||||||
|
bash build.sh
|
||||||
|
popd
|
||||||
|
fi
|
||||||
|
|
||||||
|
# input
|
||||||
|
mkdir -p data
|
||||||
|
data=$PWD/data
|
||||||
|
ckpt_dir=$data/model
|
||||||
|
model_dir=$ckpt_dir/exp/deepspeech2_online/checkpoints/
|
||||||
|
vocb_dir=$ckpt_dir/data/lang_char
|
||||||
|
# output
|
||||||
|
aishell_wav_scp=aishell_test.scp
|
||||||
|
if [ ! -d $data/test ]; then
|
||||||
|
pushd $data
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/s2t/paddle_asr_online/aishell_test.zip
|
||||||
|
unzip aishell_test.zip
|
||||||
|
popd
|
||||||
|
|
||||||
|
realpath $data/test/*/*.wav > $data/wavlist
|
||||||
|
awk -F '/' '{ print $(NF) }' $data/wavlist | awk -F '.' '{ print $1 }' > $data/utt_id
|
||||||
|
paste $data/utt_id $data/wavlist > $data/$aishell_wav_scp
|
||||||
|
fi
|
||||||
|
|
||||||
|
export GLOG_logtostderr=1
|
||||||
|
|
||||||
|
# websocket client
|
||||||
|
websocket_client_main \
|
||||||
|
--wav_rspecifier=scp:$data/$aishell_wav_scp --streaming_chunk=0.36
|
||||||
@ -0,0 +1,82 @@
|
|||||||
|
// Copyright (c) 2022 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.
|
||||||
|
// You may obtain a copy of the License at
|
||||||
|
//
|
||||||
|
// http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
//
|
||||||
|
// Unless required by applicable law or agreed to in writing, software
|
||||||
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
// See the License for the specific language governing permissions and
|
||||||
|
// limitations under the License.
|
||||||
|
|
||||||
|
#include "websocket/websocket_client.h"
|
||||||
|
#include "kaldi/feat/wave-reader.h"
|
||||||
|
#include "kaldi/util/kaldi-io.h"
|
||||||
|
#include "kaldi/util/table-types.h"
|
||||||
|
|
||||||
|
DEFINE_string(host, "127.0.0.1", "host of websocket server");
|
||||||
|
DEFINE_int32(port, 8082, "port of websocket server");
|
||||||
|
DEFINE_string(wav_rspecifier, "", "test wav scp path");
|
||||||
|
DEFINE_double(streaming_chunk, 0.1, "streaming feature chunk size");
|
||||||
|
|
||||||
|
using kaldi::int16;
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
gflags::ParseCommandLineFlags(&argc, &argv, false);
|
||||||
|
google::InitGoogleLogging(argv[0]);
|
||||||
|
|
||||||
|
kaldi::SequentialTableReader<kaldi::WaveHolder> wav_reader(
|
||||||
|
FLAGS_wav_rspecifier);
|
||||||
|
|
||||||
|
const int sample_rate = 16000;
|
||||||
|
const float streaming_chunk = FLAGS_streaming_chunk;
|
||||||
|
const int chunk_sample_size = streaming_chunk * sample_rate;
|
||||||
|
|
||||||
|
for (; !wav_reader.Done(); wav_reader.Next()) {
|
||||||
|
ppspeech::WebSocketClient client(FLAGS_host, FLAGS_port);
|
||||||
|
|
||||||
|
client.SendStartSignal();
|
||||||
|
std::string utt = wav_reader.Key();
|
||||||
|
const kaldi::WaveData& wave_data = wav_reader.Value();
|
||||||
|
CHECK_EQ(wave_data.SampFreq(), sample_rate);
|
||||||
|
|
||||||
|
int32 this_channel = 0;
|
||||||
|
kaldi::SubVector<kaldi::BaseFloat> waveform(wave_data.Data(),
|
||||||
|
this_channel);
|
||||||
|
const int tot_samples = waveform.Dim();
|
||||||
|
int sample_offset = 0;
|
||||||
|
|
||||||
|
while (sample_offset < tot_samples) {
|
||||||
|
int cur_chunk_size =
|
||||||
|
std::min(chunk_sample_size, tot_samples - sample_offset);
|
||||||
|
|
||||||
|
std::vector<int16> wav_chunk(cur_chunk_size);
|
||||||
|
for (int i = 0; i < cur_chunk_size; ++i) {
|
||||||
|
wav_chunk[i] = static_cast<int16>(waveform(sample_offset + i));
|
||||||
|
}
|
||||||
|
client.SendBinaryData(wav_chunk.data(),
|
||||||
|
wav_chunk.size() * sizeof(int16));
|
||||||
|
|
||||||
|
|
||||||
|
sample_offset += cur_chunk_size;
|
||||||
|
LOG(INFO) << "Send " << cur_chunk_size << " samples";
|
||||||
|
std::this_thread::sleep_for(
|
||||||
|
std::chrono::milliseconds(static_cast<int>(1 * 1000)));
|
||||||
|
|
||||||
|
if (cur_chunk_size < chunk_sample_size) {
|
||||||
|
client.SendEndSignal();
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
while (!client.Done()) {
|
||||||
|
}
|
||||||
|
std::string result = client.GetResult();
|
||||||
|
LOG(INFO) << "utt: " << utt << " " << result;
|
||||||
|
|
||||||
|
client.Join();
|
||||||
|
}
|
||||||
|
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
@ -0,0 +1,57 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
set +x
|
||||||
|
set -e
|
||||||
|
|
||||||
|
. path.sh
|
||||||
|
|
||||||
|
|
||||||
|
# 1. compile
|
||||||
|
if [ ! -d ${SPEECHX_EXAMPLES} ]; then
|
||||||
|
pushd ${SPEECHX_ROOT}
|
||||||
|
bash build.sh
|
||||||
|
popd
|
||||||
|
fi
|
||||||
|
|
||||||
|
# input
|
||||||
|
mkdir -p data
|
||||||
|
data=$PWD/data
|
||||||
|
ckpt_dir=$data/model
|
||||||
|
model_dir=$ckpt_dir/exp/deepspeech2_online/checkpoints/
|
||||||
|
vocb_dir=$ckpt_dir/data/lang_char/
|
||||||
|
|
||||||
|
if [ ! -f $ckpt_dir/data/mean_std.json ]; then
|
||||||
|
mkdir -p $ckpt_dir
|
||||||
|
pushd $ckpt_dir
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/s2t/aishell/asr0/asr0_deepspeech2_online_aishell_ckpt_0.2.0.model.tar.gz
|
||||||
|
tar xzfv asr0_deepspeech2_online_aishell_ckpt_0.2.0.model.tar.gz
|
||||||
|
popd
|
||||||
|
fi
|
||||||
|
|
||||||
|
export GLOG_logtostderr=1
|
||||||
|
|
||||||
|
# 3. gen cmvn
|
||||||
|
cmvn=$data/cmvn.ark
|
||||||
|
cmvn-json2kaldi --json_file=$ckpt_dir/data/mean_std.json --cmvn_write_path=$cmvn
|
||||||
|
|
||||||
|
|
||||||
|
wfst=$data/wfst/
|
||||||
|
mkdir -p $wfst
|
||||||
|
if [ ! -f $wfst/aishell_graph.zip ]; then
|
||||||
|
pushd $wfst
|
||||||
|
wget -c https://paddlespeech.bj.bcebos.com/s2t/paddle_asr_online/aishell_graph.zip
|
||||||
|
unzip aishell_graph.zip
|
||||||
|
mv aishell_graph/* $wfst
|
||||||
|
popd
|
||||||
|
fi
|
||||||
|
|
||||||
|
# 5. test websocket server
|
||||||
|
websocket_server_main \
|
||||||
|
--cmvn_file=$cmvn \
|
||||||
|
--model_path=$model_dir/avg_1.jit.pdmodel \
|
||||||
|
--streaming_chunk=0.1 \
|
||||||
|
--convert2PCM32=true \
|
||||||
|
--param_path=$model_dir/avg_1.jit.pdiparams \
|
||||||
|
--word_symbol_table=$data/wfst/words.txt \
|
||||||
|
--model_output_names=softmax_0.tmp_0,tmp_5,concat_0.tmp_0,concat_1.tmp_0 \
|
||||||
|
--graph_path=$data/wfst/TLG.fst --max_active=7500 \
|
||||||
|
--acoustic_scale=1.2
|
||||||
@ -0,0 +1,30 @@
|
|||||||
|
// Copyright (c) 2022 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.
|
||||||
|
// You may obtain a copy of the License at
|
||||||
|
//
|
||||||
|
// http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
//
|
||||||
|
// Unless required by applicable law or agreed to in writing, software
|
||||||
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
// See the License for the specific language governing permissions and
|
||||||
|
// limitations under the License.
|
||||||
|
|
||||||
|
#include "websocket/websocket_server.h"
|
||||||
|
#include "decoder/param.h"
|
||||||
|
|
||||||
|
DEFINE_int32(port, 8082, "websocket listening port");
|
||||||
|
|
||||||
|
int main(int argc, char *argv[]) {
|
||||||
|
gflags::ParseCommandLineFlags(&argc, &argv, false);
|
||||||
|
google::InitGoogleLogging(argv[0]);
|
||||||
|
|
||||||
|
ppspeech::RecognizerResource resource = ppspeech::InitRecognizerResoure();
|
||||||
|
|
||||||
|
ppspeech::WebSocketServer server(FLAGS_port, resource);
|
||||||
|
LOG(INFO) << "Listening at port " << FLAGS_port;
|
||||||
|
server.Start();
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
@ -0,0 +1,30 @@
|
|||||||
|
#!/usr/bin/env bash
|
||||||
|
|
||||||
|
set -eo pipefail
|
||||||
|
|
||||||
|
data=$1
|
||||||
|
scp=$2
|
||||||
|
split_name=$3
|
||||||
|
numsplit=$4
|
||||||
|
|
||||||
|
# save in $data/split{n}
|
||||||
|
# $scp to split
|
||||||
|
#
|
||||||
|
|
||||||
|
if [[ ! $numsplit -gt 0 ]]; then
|
||||||
|
echo "Invalid num-split argument";
|
||||||
|
exit 1;
|
||||||
|
fi
|
||||||
|
|
||||||
|
directories=$(for n in `seq $numsplit`; do echo $data/split${numsplit}/$n; done)
|
||||||
|
scp_splits=$(for n in `seq $numsplit`; do echo $data/split${numsplit}/$n/${split_name}; done)
|
||||||
|
|
||||||
|
# if this mkdir fails due to argument-list being too long, iterate.
|
||||||
|
if ! mkdir -p $directories >&/dev/null; then
|
||||||
|
for n in `seq $numsplit`; do
|
||||||
|
mkdir -p $data/split${numsplit}/$n
|
||||||
|
done
|
||||||
|
fi
|
||||||
|
|
||||||
|
echo "utils/split_scp.pl $scp $scp_splits"
|
||||||
|
utils/split_scp.pl $scp $scp_splits
|
||||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
Some files were not shown because too many files have changed in this diff Show More
Loading…
Reference in new issue