In the G158 general technique:- Providing an alternative for time-based media for audio-only content' of WCAG, it's mentioned to provide a transcript for the audio source.
In the 4th step of the testing procedure, what's the difference between the 2 of sub-points given under it?
The general technique is mentioned below:
This is using the correct format so that a screen reader can identify the difference between a transcript and a text alternative.
So for a text alternative use aria-describedby and point it at the text for the audio.
For transcripts use the <track> element. (see below for the problems with this (a))
It is possible to use aria-describedby to point to a section and then have a link within that section along the lines of 'a full transcript of the conversation is available' (linking to a copy of the transcript).
This is also considered the best way of linking it together.
(a) This tends to be used when the transcript is located on another page (and also eliminates the need for using a workaround for the fact that an audio element does not directly support WebVTT).
Because of the fact that adding a <track> element requires you to play audio through a <video> element this option is the accepted way of doing things.