# VoiceRecogMiddleware: A middleware for EFB
## Notice
**Middleware ID**: `catbaron.voice_recog`
**VoiceRecogMiddleware** is a middleware for EFB to recognize content of voice message.
This middleware is modified based on `ehForwarderBot 1.0`
- Baidu
- **Deprecated for now because my Baidu account is limited by Baidu**
- **由于我的Baidu账号不能正常登陆，因此对百度引擎的支持暂时停止。**
- API from Baidu is available for free.
- You need to get `API_KEY` and `SECRET_KEY` following
http://ai.baidu.com/docs#/ASR-API-PRO/top
- Azure Cognitive Service
- Azure Cognitive Service offers 5 audio hours per month for free.
- You need to get `Key1` and `Endpoint` from https://portal.azure.com/
- IFlyTek
- IFlyTek offers 500 queries per day fro free to new users. Additional free bundles are available from 50,000 queries/90 days to 100,000 queries/year.
- You need to get `APPID`, `APISecret` and `APIKey` from https://console.xfyun.cn/services/iat
- Tencent
- Tencent offers 15,000 queries for free per month.
- Chinese ID verification is required to use free service.
- You need to get `SecretID` and `SecretKey` from https://console.cloud.tencent.com/cam/capi
- You need to use **VoiceRecogMiddleware** on top of
[EFB](https://ehforwarderbot.readthedocs.io). Please check the
document and install EFB first.
## Dependense
* Python >= 3.6
* EFB >= 2.0.0b15
* pydub
## Install and configuration
### Install
```
pip install efb-voice_recog-middleware
```
### Enable
Register to EFB
Following [this document](https://ehforwarderbot.readthedocs.io/en/latest/getting-started.html) to edit the config file. The config file by default is `$HOME/.ehforwarderbot/profiles/default/config.yaml`. It should look like:
```yaml
master_channel: foo.demo_master
slave_channels:
- foo.demo_slave
- bar.dummy
middlewares:
- foo.other_middlewares
- catbaron.voice_recog
```
You only need to add the last line to your config file.
### Configure the middleware
The config file by default is `$HOME/.ehforwarderbot/profiles/default/catbaron.voice_recog/config.yaml`.
Please create the config file if thers is not one. Edit it as:
```yaml
speech_api:
baidu:
api_key: API_KEY
secret_key: SECRET_KEY
# supported language:
# zh, zh-x-en, en, zh-yue, zh-x-sichuan, zh-x-farfield
lang: zh
azure:
key1: KEY_1
endpoint: ENDPOINT
# supported language:
# ar-EG, ar-SA, ar-AE, ar-KW, ar-QA, ca-ES,
# da-DK, de-DE, en-AU, en-CA, en-GB, en-IN,
# en-NZ, en-US, es-ES, es-MX, fi-FI, fr-CA,
# fr-FR, gu-IN, hi-IN, it-IT, ja-JP, ko-KR,
# mr-IN, nb-NO, nl-NL, pl-PL, pt-BR, pt-PT,
# ru-RU, sv-SE, ta-IN, te-IN, zh-CN, zh-HK,
# zh-TW, th-TH, tr-TR
lang: zh
tencent:
secret_id: SECRET_ID
secret_key: SECRET_KEY
# supported language: en, zh
lang: en
iflytek:
app_id: APP_ID
api_secret: API_SECRET
api_key: APP_KEY
# supported language: zh, en
lang: en
auto: true
```
Replace the section with all-caps to your own ones.
Note that you may omit the section that you do not want to enable.
---
Turn off `auto` if you want to disable auto recognition to all voice
messages. Alternatively, you may reply <code>recog`</code> to a voice
message to recognise it.
### Restart EFB.
