Live transcription

July 2025

What’s changing

MK.IO now supports Live Transcription of audio tracks.

Who is impacted

This feature is relevant for users of MK.IO who use live event transcoding and wish to enrich their content with a new subtitle track.

Why you would use this / Why it matters

Live Transcription enhances your streaming content by generating machine-transcribed subtitles directly from the spoken words in the audio feed. This feature improves accessibility and popularity of your content, because it benefits to individuals with hearing impairments, viewers in noisy environments and non-native speakers who benefit from comprehension assistance.

Additional details

Live transcription will generate subtitles from audio tracks and deliver them in WebVTT format for HLS output and TTML format for DASH output.

This feature is available for live events using live encoding, not live passthrough.

Using a AI based live transcription in MK.IO will impact the video latency by less than 5s to offers live speech-to-text capabilities.

The full list of supported languages is available in AI workflows page.

Getting started

Enabling Live Transcription for a live event is achieved during the Live Event creation. It only applies to live encoding options and requires an AI pipeline to be selected. All it takes is to select the Predefined_ACSLiveTranscription pipeline and specify the original audio track's language using the BCP-47 format.

It is also possible to provide a list of words or phrases that are expected in the audio feed in order to improve their recognition.

Availability & rollout plan

This feature will be available in all regions starting July 2025.

Pricing

For live processing, AI workloads are billed per minute of a running live event.

AI pricing is described in detail in the pricing table