platform

Voicegain Enterprise Speech-to-Text Platform (API, Portal, etc.)

Project maintained by voicegain Hosted on GitHub Pages — Theme by mattgraham

About Voicegain Platform

Voicegain provides a Speech-to-Text Platform built around a Deep Neural Network ASR engine. Voicegain Speech-to-Text supports:

open vocabulary speech transcription (real-time and off-line) and
speech recognition (using context-free grammars). Both are accessible via Web API. In addition, the recognizer is available with MRCP interface.

Apart from bare-bones speech recognition we provide other APIs that build on top of that:

Telephone Bot API - it is a callback api suitable to building IVRs and Voicebots
Speech Analytics API

Voicegain Platform is accessible in the Cloud and can also be deployed at the Edge (on-prem Edge Computing).

This repository tracks public components of the Voicegain Platform. Things like:

Repository also provides a lot of useful code:

example code - we have examples of:
- RTP streaming for real-time transcription or recognition
- simple python scripts illustrating use of various APIs
- sample node.js web applications illustrating:
- scripts illustrating real-time transcription of Twilio Media Stream
- AWS lambda script for a Voicebot - this uses RASA and Voicegain Telephone Bot API - both node.js and python versions are available
- AWS lambda script for a Voicebot using Twilio - this is similar to the bot above but uses normal Voicegain Speech-to-Text API together with Twilio Streams - it is quite a bit more complex
- websocket streaming example in Java - it send audio over websocket and receives real-time transcript result over websocket
declarative ivr - Declarative IVR is a way to specify a complete IVR flow using a simple yaml file. The yaml file gets interpreted by a Lambda fuction and uses Voicegain Telephone Bot API to hear and talk over the phone. Included is a yam file for a simple outbound survey IVR application.
utilities:
- test-transcribe.py - takes audio files from a directory and runs it through Voicegain and Google speech-to-text - if reference transcripts are available it will report WER for both
audio-sender bootstrap bundle - this is for Live Transcription. Normally you would download it via the Web Console. Here is Zendesk help article which describes the whole process.

You can learn more about Voicegain at our main website. BTW, we are offering a generous free tier that renews each month so Signup Now.