audio-analysis-service

Audio convert/analysis service.

Features

• Read audio stream from url, convert to flac format, transcript to text with google cloud api
• Analysis text with google cloud NLP api

Prerequisites

• nodejs >= 8.10
• register google cloud account and set payment method, download your credential json.

dev

git clone [email protected]:zxdong262/audio-analysis-service.git
#or git clone https://github.com/zxdong262/audio-analysis-service.git
cd audio-analysis-service

# install dependencies
yarn

# create config
cp .sample.env .env
# then edit .env, fill your google credential path

## start local server
yarn dev

Build and Run in production env

# install pm2 first if you wanna use pm2
yarn global add pm2
# or `npm i -g pm2`

# build
yarn build

# run production server
yarn prod-server

# or use pm2
pm2 start bin/pm2.yml

Use as service

• /text-analysis?text=text for text analysis, including Sentiment, Syntax, Classification, Entity sentiment.
• /audio-url-to-text transcript audio url to text with google cloud ai.
• /audio-url-analysis for audio url analysis, will transform to flac first, then transcript to text, then use google cloud ai to anlaysis.

  it('text-analysis', function(done) {
    let text = encodeURIComponent('Hi! My name is Joanna. I will read any text you type here')
    fetch(`${base}/text-analysis?text=${text}`, {
      headers: {
        'Content-Type': 'application/json'
      }
    })
      .then(res => res.json())
      .then(res => {
        expect(!!res.text).equal(true)
        done()
      })
  })

  it('audio-url-analysis', function(done) {
    fetch(`${base}/audio-url-analysis`, {
      method: 'post',
      body: JSON.stringify({
        url: `${base}/sample.mp3`,
        //set headers for auth when needed
        //headers: {}
      }),
      headers: {
        'Content-Type': 'application/json'
      }
    })
      .then(res => res.json())
      .then(res => {
        console.log(res)
        expect(!!res.text).equal(true)
        done()
      })
  })

  it('audio-url-to-text', function(done) {
    fetch(`${base}/audio-url-to-text`, {
      method: 'post',
      body: JSON.stringify({
        url: `${base}/sample.mp3`,
        //set headers for auth when needed
        //headers: {AuthCode....}
      }),
      headers: {
        'Content-Type': 'application/json'
      }
    })
      .then(res => res.text())
      .then(res => {
        expect(res).includes('name is')
        done()
      })
  })

Use as lib

npm i audio-analysis-service

import {textAnalysis} from 'audio-analysis-service/dist/text-analysis'
import {speech2text} from 'audio-analysis-service/dist/url2text'
import {toFlac} from 'audio-analysis-service/dist/voice-to-flac'
import urlAnalysis from 'audio-analysis-service/dist/url-analysis'

Test

yarn test

License

MIT