Skip to content

Latest commit

 

History

History
165 lines (143 loc) · 7.8 KB

README_doc.md

File metadata and controls

165 lines (143 loc) · 7.8 KB

Bhashini-Translation

AI-powered language translation services for Indian languages. Easy integration and extensive language support. It's integrating with Bhashini Api. It seamlessly integrates with web frontends such as Vanilla JS, React, and Angular.
Bhashini follows ISO-639 series of language codes.

Features:

Getting Started

  1. Install the package using npm or yarn
npm install bhashini-translation
or 
yarn add bhashini-translation
  1. Import the package
import bhashini from 'bhashini-translation'
  1. set your userid, UlcaApiKey, InferenceApiKey to initialize library, to get see
  • Step 1: Go to entry point of your project
  • Step 2: call
bhashini.auth("userid", "UlcaApiKey", "InferenceApiKey")
  1. Now you are ready to use the features [I will add example here later]

get authentication details

Signup here to get authentication details
- Step 1: Fill out the registration form.

  • Step 2: Perform email authentication to enable login functionality
  • Step 3: Login using the authenticated email.
  • Step 4: Open the "My Profile" section
  • Step 5: create the API Key using Generate Button under My Profile section. [App name should use lowercase words and underscores.]
  • Step 6: press generate in api key to get credentials
  • Step 7: now copy userid, UlcaApiKey, and Inference API Key Value for Meity and pass as argument in bhashini.auth("userid", "UlcaApiKey", "InferenceApiKey") function

Code Structure

bhashini-translation-library/
├── package.json
├── jsdoc.json
├── index.js
├── README.md
├── configs/
│   ├── config.js
│   ├── payload.js
│   ├── pipelineConfig.js
│   ├── tasktypeConfig.js
│   └── verification.js
├── src/
│   ├── Utils
│   │    ├── audioUri.js
│   │    └── baseApi.js
│   ├── asr.js
│   ├── nmt.js
│   ├── tts.js
│   ├── asr_nmt.js
│   ├── nmt_tts.js
│   └── asr_nmt_tts.js
└── .gitignore

Root Directory:

  • package.json: Contains metadata and configuration for npm library, containing "name": "bhashini-translation"
  • jsdoc.json: This project uses JSDoc to generate documentation for the codebase. This file specifies various settings that control how the documentation is generated.
  • index.js: The main entry point for your library, where all functionalities are exported.

Configuration Folder:

  • configs/: This directory contains configuration-related files.
  • config.js: Contains urls for the library.
    • payload.js: Handles payload related configurations.
    • pipelineConfig.js: Handles configurations related to the pipeline.
    • tasktypeConfig.js: Contains configurations for different task types.
    • verification.js: Handles verification and authentication configurations.

Source Code Directory:

  • src/: This directory contains the source code of the library.

    • Utils/: Utility functions are placed inside this directory.

      • audioUri.js: A utility function that handles audio URIs.
      • baseApi.js: A base class that provides a common structure for API requests.

    • asr.js: Contains the code for Automatic Speech Recognition (ASR) functionality.

    • nmt.js: Contains the code for Neural Machine Translation (NMT) functionality.

    • tts.js: Contains the code for Text-to-Speech (TTS) functionality.

    • asr_nmt.js: Contains the code for a combined functionality of ASR and NMT.

    • nmt_tts.js: Contains the code for a combined functionality of NMT and TTS.

    • asr_nmt_tts.js: Contains the code for a combined functionality of ASR, NMT, and TTS.

Version Control:

  • .gitignore: A file specifying which files and directories to exclude from version control.

Notice

All functions in Bhashini return promises, so it is important to handle the promise-based output for proper asynchronous handling and retrieval of results.

How to use functionality:

ASR: Automatic Speech Recognition

converts spoken language into written text
The ASR interface provides methods to convert spoken language into text and is ideal for applications that require voice-to-text conversion.
input: source language, audioContent in base64 format
output: Text

bhashini.asr('sourceLang',"Base64")

NMT: Neural Machine Translation

translate text from one language to another indic language
The NMT interface provides methods to translate text from one language to another using advanced neural network models for high-quality translations.
input: source language, target language, Text
output: Translated text from source language to target language

bhashini.nmt('sourceLang',"targetLang", "Text")

TTS: Text to Speech

converts written text into spoken words
The TTS interface allows users to convert text into natural-sounding speech with the option to choose different voices, making it useful for applications that require generating speech output.

  • If gender is not passed, the default option will be used.

input: source language, Text, gender (male or female)
output: Audio content in Base64

bhashini.tts('sourceLang',"Text", "gender")

ASR + NMT: Speech to Text Translation

speech-to-text transcription in the target language
This interface offers a seamless integration of speech recognition and translation, allowing users to transcribe spoken language and directly translate it.
input: source language,target language, audioContent in base64 format
output: Translated text

bhashini.asr_nmt('sourceLang', "targetLang", "Base64")

NMT + TTS: Text Translation to Speech

Translates written text to another language and generates audio content in base64
This interface enables users to translate text and then convert it to speech with the option to choose different voices,, making it suitable for applications that require translated speech output.

  • If gender is not passed, the default option will be used.

input: source language,targetLang, Text, gender (male or female)
output: translated text

bhashini.nmt_tts('sourceLang', "targetLang", "Text", "gender")

ASR + NMT + TTS: Speech to Speech Translation

Converts spoken language to another Indic spoken language
This comprehensive interface provides the ability to perform speech recognition, translation, and text-to-speech conversion in one call supporting voice selection, catering to complex language processing needs.

  • If gender is not passed, the default option will be used.

input: source language, target language, audioContent in base64 format, gender (male or female)
output: audio content in base64

bhashini.asr_nmt_tts('sourceLang', "targetLang", "Base64", "gender")