Audio-only material has no substitute but a transcript

What it is

Media with no picture — a podcast, audio news, a long spoken notice — has no visual surface on which to stack captions. For anyone who cannot hear that soundtrack, the alternative is a transcript. This is not “prepare several media for deaf users, any of them will do.” It is the required channel on an audio-only product line. Captions parasitise video. There is no video here.

Why it happens

Closed captions need a clock and a rendering surface. An audio player is usually a scrubber and a cover image, with no caption layer. Even when the listener can hear, libraries, carriages and late nights often forbid sound; reading is then the only channel. Pre-recorded audio without text keeps the proposition in the waveform alone. Close that channel and nothing remains.

Video has its own caption rule. That rule does not retire the podcast transcript, and the transcript does not retire captions on video — those users are on another medium. Audio-only fails more cleanly: there is no lip, no slide, to glance at. Either the text exists, or the content does not.

Studying it

Audit before you survey. Of every audio object on the site (in-page audio, podcast RSS, in-app voice notes), how many ship independently readable text? Then unplug the speakers and ask “what three things did this episode say,” using only the words on the page.

Independent variables: transcript present or not, in-body versus a download, whether the download is readable by assistive technology. Dependent variables: whether the points can be reported in silence, whether the report matches the audio’s propositions.

The sample must include objects that “do not look like video, so nobody thought of captions.” Otherwise the audit only sees course recordings.

Where it stops holding

A button click that carries no proposition is not audio-only content and does not need a transcript. A live call’s after-the-fact notes raise consent and delay issues that a pre-recorded podcast checklist cannot swallow. Sound art with no linguistic proposition wants a production note, not a dialogue equivalent. A spoken prompt whose words already sit in full on the same screen is a duplicate if reattached as a transcript.

Applying it

• Ship every podcast episode and audio lesson with full text in the body of the play page, not only as a file behind another click.
• On the release checklist, list audio objects as their own row. Do not only inspect picture files for a caption track.
• How to check: unplug the sound. Answer “what three things did this episode say” from the page alone. If any step requires pressing play to learn the proposition, that audio still has no alternative channel.

Related

• Same group: J4.02.1 A transcript is there to be searched and skimmed · J4.02.3 Each line of the transcript must sit on the media clock
• Nearby: J4.01 Captions · A3.08 Hearing loss implications for interaction design · J1.01 Four Principles
• Search terms: audio-only transcript · podcast transcript · media alternative