You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
PaddleSpeech/demos/speech_server/README_cn.md

430 lines
21 KiB

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

(简体中文|[English](./README.md))
# 语音服务
## 介绍
这个 demo 是一个启动离线语音服务和访问服务的实现。它可以通过使用 `paddlespeech_server``paddlespeech_client` 的单个命令或 python 的几行代码来实现。
服务接口定义请参考:
- [PaddleSpeech Server RESTful API](https://github.com/PaddlePaddle/PaddleSpeech/wiki/PaddleSpeech-Server-RESTful-API)
## 使用方法
### 1. 安装
请看 [安装文档](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/docs/source/install.md).
推荐使用 **paddlepaddle 2.3.1** 或以上版本。
你可以从简单,中等,困难 几种方式中选择一种方式安装 PaddleSpeech。
**如果使用简单模式安装,需要自行准备 yaml 文件,可参考 conf 目录下的 yaml 文件。**
### 2. 准备配置文件
配置文件可参见 `conf/application.yaml`
其中,`engine_list` 表示即将启动的服务将会包含的语音引擎,格式为 <语音任务>_<引擎类型>。
目前服务集成的语音任务有: asr (语音识别)、tts (语音合成)、cls (音频分类)、vector (声纹识别)以及 text (文本处理)。
目前引擎类型支持两种形式python 及 inference (Paddle Inference)
**注意:** 如果在容器里可正常启动服务,但客户端访问 ip 不可达,可尝试将配置文件中 `host` 地址换成本地 ip 地址。
### 3. 服务端使用方法
- 命令行 (推荐使用)
```bash
# 启动服务
paddlespeech_server start --config_file ./conf/application.yaml
```
使用方法:
```bash
paddlespeech_server start --help
```
参数:
- `config_file`: 服务的配置文件,默认: ./conf/application.yaml
- `log_file`: log 文件. 默认:./log/paddlespeech.log
输出:
```text
[2022-02-23 11:17:32] [INFO] [server.py:64] Started server process [6384]
INFO: Waiting for application startup.
[2022-02-23 11:17:32] [INFO] [on.py:26] Waiting for application startup.
INFO: Application startup complete.
[2022-02-23 11:17:32] [INFO] [on.py:38] Application startup complete.
INFO: Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
[2022-02-23 11:17:32] [INFO] [server.py:204] 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/application.yaml",
log_file="./log/paddlespeech.log")
```
输出:
```text
INFO: Started server process [529]
[2022-02-23 14:57:56] [INFO] [server.py:64] Started server process [529]
INFO: Waiting for application startup.
[2022-02-23 14:57:56] [INFO] [on.py:26] Waiting for application startup.
INFO: Application startup complete.
[2022-02-23 14:57:56] [INFO] [on.py:38] Application startup complete.
INFO: Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
[2022-02-23 14:57:56] [INFO] [server.py:204] Uvicorn running on http://0.0.0.0:8090 (Press CTRL+C to quit)
```
### 4. ASR 客户端使用方法
ASR 客户端的输入是一个 WAV 文件(`.wav`),并且采样率必须与模型的采样率相同。
可以下载 ASR 客户端的示例音频:
```bash
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav
```
**注意:** 初次使用客户端时响应时间会略长
- 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
```
paddlespeech_client asr --server_ip 127.0.0.1 --port 8090 --input ./zh.wav
```
使用帮助:
```bash
paddlespeech_client asr --help
```
参数:
- `server_ip`: 服务端 ip 地址,默认: 127.0.0.1。
- `port`: 服务端口,默认: 8090。
- `input`(必须输入): 用于识别的音频文件。
- `sample_rate`: 音频采样率默认值16000。
- `lang`: 模型语言默认值zh_cn。
- `audio_format`: 音频格式默认值wav。
输出:
```text
[2022-08-01 07:54:01,646] [ INFO] - ASR result: 我认为跑步最重要的就是给我带来了身体健康
[2022-08-01 07:54:01,646] [ INFO] - Response time 4.898965 s.
```
- Python API
```python
from paddlespeech.server.bin.paddlespeech_client import ASRClientExecutor
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)
```
输出:
```text
我认为跑步最重要的就是给我带来了身体健康
```
### 5. TTS 客户端使用方法
**注意:** 初次使用客户端时响应时间会略长
- 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
```bash
paddlespeech_client tts --server_ip 127.0.0.1 --port 8090 --input "您好,欢迎使用百度飞桨语音合成服务。" --output output.wav
```
使用帮助:
```bash
paddlespeech_client tts --help
```
参数:
- `server_ip`: 服务端ip地址默认: 127.0.0.1。
- `port`: 服务端口,默认: 8090。
- `input`(必须输入): 待合成的文本。
- `spk_id`: 说话人 id用于多说话人语音合成默认值 0。
- `speed`: 音频速度,该值应设置在 0 到 3 之间。 默认值1.0
- `volume`: 音频音量,该值应设置在 0 到 3 之间。 默认值: 1.0
- `sample_rate`: 采样率,可选 [0, 8000, 16000],默认与模型相同。 默认值0
- `output`: 输出音频的路径, 默认值None表示不保存音频到本地。
输出:
```text
[2022-02-23 15:20:37,875] [ INFO] - Save synthesized audio successfully on output.wav.
[2022-02-23 15:20:37,875] [ INFO] - Audio duration: 3.612500 s.
[2022-02-23 15:20:37,875] [ INFO] - Response time: 0.348050 s.
```
- Python API
```python
from paddlespeech.server.bin.paddlespeech_client import TTSClientExecutor
import json
ttsclient_executor = TTSClientExecutor()
res = ttsclient_executor(
input="您好,欢迎使用百度飞桨语音合成服务。",
server_ip="127.0.0.1",
port=8090,
spk_id=0,
speed=1.0,
volume=1.0,
sample_rate=0,
output="./output.wav")
response_dict = res.json()
print(response_dict["message"])
print("Save synthesized audio successfully on %s." % (response_dict['result']['save_path']))
print("Audio duration: %f s." %(response_dict['result']['duration']))
```
输出:
```text
{'description': 'success.'}
Save synthesized audio successfully on ./output.wav.
Audio duration: 3.612500 s.
```
### 6. CLS 客户端使用方法
可以下载 CLS 客户端的示例音频:
```bash
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
```
**注意:** 初次使用客户端时响应时间会略长
- 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
```bash
paddlespeech_client cls --server_ip 127.0.0.1 --port 8090 --input ./zh.wav
```
使用帮助:
```bash
paddlespeech_client cls --help
```
参数:
- `server_ip`: 服务端 ip 地址,默认: 127.0.0.1。
- `port`: 服务端口,默认: 8090。
- `input`(必须输入): 用于分类的音频文件。
- `topk`: 分类结果的topk。
输出:
```text
[2022-03-09 20:44:39,974] [ INFO] - {'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'topk': 1, 'results': [{'class_name': 'Speech', 'prob': 0.9027184844017029}]}}
[2022-03-09 20:44:39,975] [ INFO] - Response time 0.104360 s.
```
- Python API
```python
from paddlespeech.server.bin.paddlespeech_client import CLSClientExecutor
import json
clsclient_executor = CLSClientExecutor()
res = clsclient_executor(
input="./zh.wav",
server_ip="127.0.0.1",
port=8090,
topk=1)
print(res.json())
```
输出:
```text
{'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'topk': 1, 'results': [{'class_name': 'Speech', 'prob': 0.9027184844017029}]}}
```
### 7. 声纹客户端使用方法
可以下载声纹客户端的示例音频:
```bash
wget -c https://paddlespeech.bj.bcebos.com/vector/audio/85236145389.wav
wget -c https://paddlespeech.bj.bcebos.com/vector/audio/123456789.wav
```
#### 7.1 提取声纹特征
**注意:** 初次使用客户端时响应时间会略长
* 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
```bash
paddlespeech_client vector --task spk --server_ip 127.0.0.1 --port 8090 --input 85236145389.wav
```
使用帮助:
``` bash
paddlespeech_client vector --help
```
参数:
* server_ip: 服务端ip地址默认: 127.0.0.1。
* port: 服务端口,默认: 8090。
* input(必须输入): 用于识别的音频文件。
* task: vector 的任务可选spk或者score。默认是 spk。
* enroll: 注册音频;。
* test: 测试音频。
输出:
```text
[2022-08-01 09:01:22,151] [ INFO] - vector http client start
[2022-08-01 09:01:22,152] [ INFO] - the input audio: 85236145389.wav
[2022-08-01 09:01:22,152] [ INFO] - endpoint: http://127.0.0.1:8090/paddlespeech/vector
[2022-08-01 09:01:27,093] [ INFO] - {'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'vec': [1.4217487573623657, 5.626248836517334, -5.342073440551758, 1.177390217781067, 3.308061122894287, 1.7565997838974, 5.1678876876831055, 10.806346893310547, -3.822679042816162, -5.614130973815918, 2.6238481998443604, -0.8072965741157532, 1.963512659072876, -7.312864780426025, 0.011034967377781868, -9.723127365112305, 0.661963164806366, -6.976816654205322, 10.213465690612793, 7.494767189025879, 2.9105641841888428, 3.894925117492676, 3.7999846935272217, 7.106173992156982, 16.905324935913086, -7.149376392364502, 8.733112335205078, 3.423002004623413, -4.831653118133545, -11.403371810913086, 11.232216835021973, 7.127464771270752, -4.282831192016602, 2.4523589611053467, -5.13075065612793, -18.17765998840332, -2.611666440963745, -11.00034236907959, -6.731431007385254, 1.6564655303955078, 0.7618184685707092, 1.1253058910369873, -2.0838277339935303, 4.725739002227783, -8.782590866088867, -3.5398736000061035, 3.8142387866973877, 5.142062664031982, 2.162053346633911, 4.09642219543457, -6.416221618652344, 12.747454643249512, 1.9429889917373657, -15.152948379516602, 6.417416572570801, 16.097013473510742, -9.716649055480957, -1.9920448064804077, -3.364956855773926, -1.8719490766525269, 11.567351341247559, 3.6978795528411865, 11.258269309997559, 7.442364692687988, 9.183405876159668, 4.528151512145996, -1.2417811155319214, 4.395910263061523, 6.672768592834473, 5.889888763427734, 7.627115249633789, -0.6692016124725342, -11.889703750610352, -9.208883285522461, -7.427401542663574, -3.777655601501465, 6.917237758636475, -9.848749160766602, -2.094479560852051, -5.1351189613342285, 0.49564215540885925, 9.317541122436523, -5.9141845703125, -1.809845209121704, -0.11738205701112747, -7.169270992279053, -1.0578246116638184, -5.721685886383057, -5.117387294769287, 16.137670516967773, -4.473618984222412, 7.66243314743042, -0.5538089871406555, 9.631582260131836, -6.470466613769531, -8.54850959777832, 4.371622085571289, -0.7970349192619324, 4.479003429412842, -2.9758646488189697, 3.2721707820892334, 2.8382749557495117, 5.1345953941345215, -9.19078254699707, -0.5657423138618469, -4.874573230743408, 2.316561460494995, -5.984307289123535, -2.1798791885375977, 0.35541653633117676, -0.3178458511829376, 9.493547439575195, 2.114448070526123, 4.358088493347168, -12.089820861816406, 8.451695442199707, -7.925461769104004, 4.624246120452881, 4.428938388824463, 18.691999435424805, -2.620460033416748, -5.149182319641113, -0.3582168221473694, 8.488557815551758, 4.98148250579834, -9.326834678649902, -2.2544236183166504, 6.64176607131958, 1.2119656801223755, 10.977132797241211, 16.55504035949707, 3.323848247528076, 9.55185317993164, -1.6677050590515137, -0.7953923940658569, -8.605660438537598, -0.4735637903213501, 2.6741855144500732, -5.359188079833984, -2.6673784255981445, 0.6660736799240112, 15.443212509155273, 4.740597724914551, -3.4725306034088135, 11.592561721801758, -2.05450701713562, 1.7361239194869995, -8.26533031463623, -9.304476737976074, 5.406835079193115, -1.5180232524871826, -7.746610641479492, -6.089605331420898, 0.07112561166286469, -0.34904858469963074, -8.649889945983887, -9.998958587646484, -2.5648481845855713, -0.5399898886680603, 2.6018145084381104, -0.31927648186683655, -1.8815231323242188, -2.0721378326416016, -3.4105639457702637, -8.299802780151367, 1.4836379289627075, -15.366002082824707, -8.288193702697754, 3.884773015975952, -3.4876506328582764, 7.362995624542236, 0.4657321572303772, 3.1326000690460205, 12.438883781433105, -1.8337029218673706, 4.532927513122559, 2.726433277130127, 10.145345687866211, -6.521956920623779, 2.8971481323242188, -3.3925881385803223, 5.079156398773193, 7.759725093841553, 4.677562236785889, 5.8457818031311035, 2.4023921489715576, 7.707108974456787, 3.9711389541625977, -6.390035152435303, 6.126871109008789, -3.776031017303467, -11.118141174316406]}}
[2022-08-01 09:01:27,094] [ INFO] - Response time 4.941739 s.
```
* Python API
``` python
from paddlespeech.server.bin.paddlespeech_client import VectorClientExecutor
import json
vectorclient_executor = VectorClientExecutor()
res = vectorclient_executor(
input="85236145389.wav",
server_ip="127.0.0.1",
port=8090,
task="spk")
print(res.json())
```
输出:
```text
{'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'vec': [1.4217487573623657, 5.626248836517334, -5.342073440551758, 1.177390217781067, 3.308061122894287, 1.7565997838974, 5.1678876876831055, 10.806346893310547, -3.822679042816162, -5.614130973815918, 2.6238481998443604, -0.8072965741157532, 1.963512659072876, -7.312864780426025, 0.011034967377781868, -9.723127365112305, 0.661963164806366, -6.976816654205322, 10.213465690612793, 7.494767189025879, 2.9105641841888428, 3.894925117492676, 3.7999846935272217, 7.106173992156982, 16.905324935913086, -7.149376392364502, 8.733112335205078, 3.423002004623413, -4.831653118133545, -11.403371810913086, 11.232216835021973, 7.127464771270752, -4.282831192016602, 2.4523589611053467, -5.13075065612793, -18.17765998840332, -2.611666440963745, -11.00034236907959, -6.731431007385254, 1.6564655303955078, 0.7618184685707092, 1.1253058910369873, -2.0838277339935303, 4.725739002227783, -8.782590866088867, -3.5398736000061035, 3.8142387866973877, 5.142062664031982, 2.162053346633911, 4.09642219543457, -6.416221618652344, 12.747454643249512, 1.9429889917373657, -15.152948379516602, 6.417416572570801, 16.097013473510742, -9.716649055480957, -1.9920448064804077, -3.364956855773926, -1.8719490766525269, 11.567351341247559, 3.6978795528411865, 11.258269309997559, 7.442364692687988, 9.183405876159668, 4.528151512145996, -1.2417811155319214, 4.395910263061523, 6.672768592834473, 5.889888763427734, 7.627115249633789, -0.6692016124725342, -11.889703750610352, -9.208883285522461, -7.427401542663574, -3.777655601501465, 6.917237758636475, -9.848749160766602, -2.094479560852051, -5.1351189613342285, 0.49564215540885925, 9.317541122436523, -5.9141845703125, -1.809845209121704, -0.11738205701112747, -7.169270992279053, -1.0578246116638184, -5.721685886383057, -5.117387294769287, 16.137670516967773, -4.473618984222412, 7.66243314743042, -0.5538089871406555, 9.631582260131836, -6.470466613769531, -8.54850959777832, 4.371622085571289, -0.7970349192619324, 4.479003429412842, -2.9758646488189697, 3.2721707820892334, 2.8382749557495117, 5.1345953941345215, -9.19078254699707, -0.5657423138618469, -4.874573230743408, 2.316561460494995, -5.984307289123535, -2.1798791885375977, 0.35541653633117676, -0.3178458511829376, 9.493547439575195, 2.114448070526123, 4.358088493347168, -12.089820861816406, 8.451695442199707, -7.925461769104004, 4.624246120452881, 4.428938388824463, 18.691999435424805, -2.620460033416748, -5.149182319641113, -0.3582168221473694, 8.488557815551758, 4.98148250579834, -9.326834678649902, -2.2544236183166504, 6.64176607131958, 1.2119656801223755, 10.977132797241211, 16.55504035949707, 3.323848247528076, 9.55185317993164, -1.6677050590515137, -0.7953923940658569, -8.605660438537598, -0.4735637903213501, 2.6741855144500732, -5.359188079833984, -2.6673784255981445, 0.6660736799240112, 15.443212509155273, 4.740597724914551, -3.4725306034088135, 11.592561721801758, -2.05450701713562, 1.7361239194869995, -8.26533031463623, -9.304476737976074, 5.406835079193115, -1.5180232524871826, -7.746610641479492, -6.089605331420898, 0.07112561166286469, -0.34904858469963074, -8.649889945983887, -9.998958587646484, -2.5648481845855713, -0.5399898886680603, 2.6018145084381104, -0.31927648186683655, -1.8815231323242188, -2.0721378326416016, -3.4105639457702637, -8.299802780151367, 1.4836379289627075, -15.366002082824707, -8.288193702697754, 3.884773015975952, -3.4876506328582764, 7.362995624542236, 0.4657321572303772, 3.1326000690460205, 12.438883781433105, -1.8337029218673706, 4.532927513122559, 2.726433277130127, 10.145345687866211, -6.521956920623779, 2.8971481323242188, -3.3925881385803223, 5.079156398773193, 7.759725093841553, 4.677562236785889, 5.8457818031311035, 2.4023921489715576, 7.707108974456787, 3.9711389541625977, -6.390035152435303, 6.126871109008789, -3.776031017303467, -11.118141174316406]}}
```
#### 7.2 音频声纹打分
**注意:** 初次使用客户端时响应时间会略长
* 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
``` bash
paddlespeech_client vector --task score --server_ip 127.0.0.1 --port 8090 --enroll 85236145389.wav --test 123456789.wav
```
使用帮助:
``` bash
paddlespeech_client vector --help
```
参数:
* server_ip: 服务端ip地址默认: 127.0.0.1。
* port: 服务端口,默认: 8090。
* input(必须输入): 用于识别的音频文件。
* task: vector 的任务可选spk或者score。默认是 spk。
* enroll: 注册音频;。
* test: 测试音频。
输出:
```text
[2022-08-01 09:04:42,275] [ INFO] - vector score http client start
[2022-08-01 09:04:42,275] [ INFO] - enroll audio: 85236145389.wav, test audio: 123456789.wav
[2022-08-01 09:04:42,275] [ INFO] - endpoint: http://127.0.0.1:8090/paddlespeech/vector/score
[2022-08-01 09:04:44,611] [ INFO] - {'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'score': 0.4292638897895813}}
[2022-08-01 09:04:44,611] [ INFO] - Response time 2.336258 s.
```
* Python API
```python
from paddlespeech.server.bin.paddlespeech_client import VectorClientExecutor
import json
vectorclient_executor = VectorClientExecutor()
res = vectorclient_executor(
input=None,
enroll_audio="85236145389.wav",
test_audio="123456789.wav",
server_ip="127.0.0.1",
port=8090,
task="score")
print(res.json())
```
输出:
```text
{'success': True, 'code': 200, 'message': {'description': 'success'}, 'result': {'score': 0.4292638897895813}}
```
### 8. 标点预测
**注意:** 初次使用客户端时响应时间会略长
- 命令行 (推荐使用)
`127.0.0.1` 不能访问,则需要使用实际服务 IP 地址
``` bash
paddlespeech_client text --server_ip 127.0.0.1 --port 8090 --input "我认为跑步最重要的就是给我带来了身体健康"
```
使用帮助:
```bash
paddlespeech_client text --help
```
参数:
- `server_ip`: 服务端ip地址默认: 127.0.0.1。
- `port`: 服务端口,默认: 8090。
- `input`(必须输入): 用于标点预测的文本内容。
输出:
```text
[2022-05-09 18:19:04,397] [ INFO] - The punc text: 我认为跑步最重要的就是给我带来了身体健康。
[2022-05-09 18:19:04,397] [ INFO] - Response time 0.092407 s.
```
- Python API
```python
from paddlespeech.server.bin.paddlespeech_client import TextClientExecutor
textclient_executor = TextClientExecutor()
res = textclient_executor(
input="我认为跑步最重要的就是给我带来了身体健康",
server_ip="127.0.0.1",
port=8090,)
print(res)
```
输出:
```text
我认为跑步最重要的就是给我带来了身体健康。
```
## 服务支持的模型
### ASR 支持的模型
通过 `paddlespeech_server stats --task asr` 获取 ASR 服务支持的所有模型,其中静态模型可用于 paddle inference 推理。
### TTS 支持的模型
通过 `paddlespeech_server stats --task tts` 获取 TTS 服务支持的所有模型,其中静态模型可用于 paddle inference 推理。
### CLS 支持的模型
通过 `paddlespeech_server stats --task cls` 获取 CLS 服务支持的所有模型,其中静态模型可用于 paddle inference 推理。
### Vector 支持的模型
通过 `paddlespeech_server stats --task vector` 获取 Vector 服务支持的所有模型。
### Text支持的模型
通过 `paddlespeech_server stats --task text` 获取 Text 服务支持的所有模型。