Inference Client Website for ASR(Automatic Speech Recognition) & TTS(Text To Speech) models

Summary

This website supports realtime speech to text processing which can be connected to the speech recognition open api provided here.

A feedback component is added where for every streaming session(Interval between switching on the mic to switching it off), feedback can be collected through star ratings.

This website also supports SRT(SubRip Subtitle) file generation.

It also supports text to speech processing in the url(/tts).

Getting Started Guide

1. Download and Install nodejs and npm latest stable version.

2. Install the node dependencies for this project using the following command:npm i.

3. Update the list of languages you need to support in the language_details.json file in project root folder. eg:

   [
       {"id": "indian-english", "code": "en", "name": "English (IN)"},
       {"id":"hindi", "code": "hi", "name":"Hindi"}
   ]

4. Set the following environment variables for the realtime speech recognition and SRT generation.

   • OPEN_API_PROXY_URL
   • PUNCTUATION_BASE_URL
   • ASR_BASE_URL
   • TTS_BASE_URL

   eg:

   OPEN_API_PROXY_URL=https://open-api-proxy-example.com
   PUNCTUATION_BASE_URL=https://punctuation-example.com
   ASR_BASE_URL=https://asr-base-url-example.com
   TTS_BASE_URL=https://tts-base-url-example.com

   Note: The environment variables can be set by creating a .env file in the project root folder and the above environment variables can be added in that file.

5. If feedback feature is needed, follow these steps:

   • Create a database table with the following columns:

   • Create a gcp bucket where the audio processed can be stored in for the given feedback.

   • Set Google application credentials in the development/deployment environment to get access to gcp buckets Or you can get credentials.json file for the gcp account and provide the path of the file to the environment variable:GOOGLE_APPLICATION_CREDENTIALS.

   • Add the following environment variables:

     DB_USER=username
     DB_PASS=password
     DB_NAME=dbname
     DB_HOST=dbhost
     BUCKET_NAME=bucketname_to_store_audios
     GOOGLE_APPLICATION_CREDENTIALS=<credentials.json file path>

6. Run the following command to get the website up and running: npm start.

7. The website will be running in port 9008.

Supported environment variables

Variable Name	Description
OPEN_API_PROXY_URL	This url points to the proxy server which makes the streaming possible.
PUNCTUATION_BASE_URL	This url points to the punctuation api.
ASR_BASE_URL	URL where ASR models are hosted through REST APIs
TTS_BASE_URL	URL where TTS models are hosted through REST APIs
DB_USER	Database username
DB_PASS	Database password
DB_NAME	Database name
DB_HOST	Database host/ip
BUCKET_NAME	Bucket name in google cloud to store audios
GOOGLE_APPLICATION_CREDENTIALS	Google application credentials file path to access the gcp bucket.