Skip to content

Latest commit

 

History

History
61 lines (45 loc) · 1.56 KB

README.md

File metadata and controls

61 lines (45 loc) · 1.56 KB
Raw

Vietnamese ASR + Google Translations

Introduction

This simple API is to combine Automatic Speech Recognition for Vietnamese and then translate to the other supported langagues from Google Translations client library. I utilized FastAPI for building this (Great opensource for fast and efficient building the API applications).

Vietnamese ASR

Please refer to the opensource https://github.com/vietai/ASR

Google Translations client

Please refer to the opensource https://pypi.org/project/googletrans/

Web framework (FastAPI)

Please refer to the opensource https://fastapi.tiangolo.com/

Web UI

Pull source code for setup

git pull https://github.com/hoangtuanvu/vietnamese_asr_service.git
cd vietnamese_asr_service

Pretrained ASR model

Please download follow by the instruction of the source link (HuggingFace) and then copy all relevant directories and files to 'app/pretrained_models' folder.

Build Docker Image

docker build -t vi_asr_image .

Run API

docker run -d --name vi_asr -p 8000:80 vi_asr_image:latest

or do it locally

uvicorn app.service:app --host 0.0.0.0 --port 8000

Mockup API testing

pip install pytest
pytest app/test_service.py

Note: Please note that, there are some common use-cases. You then add more if needed.

Use WebUI

Please open the WebUI followed by

http://host:port/docs

And then fill data input such as 'file' and 'data'.