Below we present all components involved in live (real-time) transcription.

Functionality

Live Transcription takes audio input, transcribes it in real-time, and then broadcasts the transcript to multiple viewers via the web based CC-App. It can be used without any programming, but also provides access to a complete Web API which allows for building own custom live transcription applications.

Live Transcribe is mainly intended to be used for live events where the transcript needs to be available to the viewers (e.g. people hearing impaired) as close to the real time as possible. Voicegain Live Transcribe feature has provisions for multiple audio streams being transcribed and controlled at the same time which makes is suitable for large events like conferences, or for use in education (transcription in multiple classrooms controlled at the same time).

Components

The diagram shows the major components involved in Live Transcription setup.

The components are:

• Voicegain Platform hosted in the Cloud or On-Prem (at the Edge) - it contains the Speech Recognition engine, Web API services, and storage.
• Web Browser for managing Live Transcription - this is used to:
  • define the broadcast websockets
  • manage transcription settings, e.g., Language Models
  • control the Audio-Sender Daemon(s)
• Audio-Sender Daemon - reads data from an audio device (e.g. USB sound device) and streams the audio data to the Speech-to-Text engine on the Voicegain Platform. The daemon may be controlled remotely via a web browser - using start, stop, mute, unmute commands.
• CC-App - this a web browser based viewer for the live transcript. It receives the transcript send via broadcast websockets.
• Web Browser for access to archives - end-users have access to the audio and transcripts after the session finishes. It also shows which broadcast websockets a given account has publicly available.

Broadcast Websockets

Live Transcripts are distributed using Broadcast Websockets. In order for them to be accessed from the CC-App, the websockets need to be first defined. This can be done from the Voicegain Web Console or via an API. For more about creating and managing Broadcast Websockets see here.

CC-App viewer

Voicegain provides a standard viewer for live transcripts broadcast using the websockets. The viewer is launched from this url https://cc.voicegain.ai/<account-web-name> where <account-web-name> is a web friendly name of your Voicegain account .

This page will show all public broadcast websockets on an account, their status, and the Launch button.

You can also go directly to the page for a specific websocket, e.g., given the websockets above we can go directly to https://cc.voicegain.ai/May13 - notice that it is not needed to include the <account-web-name> in this url.

In the video below you can see an example of the CC-App in action: