# ASR 安裝 筆記
wiki:
http://wiki.emotibot.com:8090/pages/viewpage.action?pageId=18645557

http://192.168.3.233:9000/minio/login
http://192.168.3.222:9000/minio/login
aicctp1.emotibot.com

voice_deploy
### Config setting
1. DEPLOY_MINIO => True change to False (若已安裝，因此不需要部署)
2. ASR_MODEL_NAME => set "" (台北Module使用此Module)
1. tag: v0.3.5.1 => tp_asr8k_v13.0.6_tra_v4
2. ??? => tp_asr8k_v13.0.5_tra_v4

### install
3. ./scripts/deploy.sh deploy
4. 上傳 RESCORE Module 至相對應目錄 (目錄為建立則自行建立)
1. 放小模型到 minio
1.可用browser UI http://192.168.3.215:9000/
2.或手動放到此目錄下：~/voice/asr_models/{ASR_MODEL_NAME}/rescore/emotibot/{rescore_model_name}
2. 戳一下這個api 才會正式啟動
curl -XPOST http://localhost:18003/rescore/api/v1/deploy/{小模型名稱}
### check
4. 確認相關資訊
curl -XGET http://localhost:18000/status | python -m json.tool
5. 測試ASR 是否成功啟動
docker exec -it voice_master_emotibot-deployer /test.sh recognize
6. 音檔測試 (音訊 自行尋找) 執行完會分析出來音訊的文字
curl -X POST -T ~/voice_deploy/test/test_8k.wav "http://localhost:18000/client/dynamic/recognize?custom_rescore=4b21158a395311e88a710242ac110002_model_1"

curl -X POST -T "http://localhost:18000/client/dynamic/recognize"

打包 image:
./scripts/deploy.sh package