Publish a full text transcript beside every podcast, interview and recorded talk on your site. WCAG 2.0 asks for one at its most basic level, and the same text lets search engines and skimmers find what was said.

Audio with nothing to read

A business starts a podcast, records a talk, or interviews a long-standing customer. The file goes on the website with a play button and a two-line summary.

For a deaf visitor, that page is a title and a button that does nothing for them. For someone hard of hearing, a long conversation recorded in a busy room is hard work.

It also fails people who could listen and will not: the reader at an open-plan desk without headphones, the parent beside a sleeping baby, the visitor who reads faster than anyone talks.

And it gives search engines almost nothing, because they read the words on a page and this page has very few.

Who a missing transcript leaves out

The fourth is the one owners underestimate. Few people will play a forty-minute episode to find out whether it covers their question, but most will scan a page of text for it.

What the standard asks for

WCAG 2.0, the guidelines most accessibility requirements point to, covers recorded audio at Level A, its most basic level.

For audio-only content such as a podcast or a recorded talk, it asks for a text alternative that presents the same information. In practice, that is a transcript.

A summary does not meet it, and neither does a list of topics with timings. The test is whether someone who cannot hear gets everything a listener gets.

Video with sound has separate requirements, starting with captions.

What a good transcript contains

Every word that carries meaning, with each speaker named the first time and marked at every change, so a reader can follow a conversation between three people.

Light editing is fine. Removing ums, false starts and repeated words makes the text easier to read without changing what anyone said.

Sounds that matter go in brackets: laughter that changes the sense of a remark, a clip being discussed, music that opens a segment.

And names spelled correctly: guests, products, streets, local places. Those are the words people search for, and the words automatic tools get wrong.

Where it goes on the page

Put the transcript on the same page as the player, as ordinary text: in full below it, or behind a clearly labelled link right beside it.

Avoid a PDF, which is often unreadable to screen readers when exported carelessly and awkward on a phone. Never post an image of text, which nothing can read aloud or index.

The player should work from a keyboard alone, and audio that starts by itself needs a way to stop it, though letting visitors press play is better.

A worked example

A comic book shop in a suburban plaza records a weekly podcast. Two staff members talk through the new releases arriving that Wednesday, and once a month they interview a local artist.

Each episode went on the shop's website as a player, an episode number and one line listing the titles discussed.

A regular customer who is deaf asked at the counter whether there was a written version. The staff picks and the advice on what to add to a pull list lived only in the recordings.

To search engines the episode pages were nearly empty, so a series the hosts discussed for twenty minutes never led anyone to the shop.

The shop now runs each recording through an automatic transcription service, and one host spends about an hour correcting it: character names, creators, series titles, the artist's studio.

The edited transcript goes below the player, with a heading for each title discussed. The customer reads every episode now.

Producing it without a week of typing

Typing an hour of conversation by hand takes several hours, and most businesses should not try.

Automatic transcription is now quick and inexpensive, and gives a usable first draft. It stumbles on names, accents, crosstalk and anything said over music, so it needs a careful edit by someone who knows the subject.

Professional transcription costs more and is more accurate. For a talk or interview that will stay on the site for years, that is usually money well spent.

Either way, build the edit into the schedule. A transcript promised for later rarely arrives.

What the text does for search

Search engines index the words on a page. An episode page with a title, a player and two lines gives them two lines.

A transcript gives them every product, place, problem and question the episode covered, in the words customers actually use.

That does not guarantee rankings, and a raw wall of text is a poor landing page; a short summary and headings fix that.

It also hands the business written material it already owns: quotes for social posts, the basis of a blog article, answers for a questions page.

The recordings already online

Most sites with audio have a back catalogue, and transcribing all of it at once is rarely realistic.

Start with what people still open: the most played episodes, talks linked from service pages, and anything referred to elsewhere on the site.

Some old recordings are better removed than transcribed, if what they say is out of date. Work through the rest a few at a time.

The counter-case

Audio that reads out an article already on the page is itself the alternative, and the text is already there. The guidelines make an exception for exactly that case.

A short sound with no information in it, such as a jingle, needs a line saying what it is rather than a transcript.

And a business with years of recordings can reasonably catch up over months, as long as the order is deliberate and new recordings come first.

For everything else, the audio and its transcript should go up together.

Before the next recording

  1. List every recording on the site.
  2. Check which ones people still play.
  3. Transcribe the next new one before it goes live.
  4. Have someone who knows the subject edit the names.
  5. Put the text on the page, not in a PDF.
  6. Add headings at each change of topic.
  7. Work back through older recordings, most played first.

Step three is what stops the backlog growing. Old recordings can be caught up over months, but every new one published without text adds another page to fix.

Another accessibility check that needs nothing more than a keyboard is in tab order that jumps around the page.

Frequently asked questions

Why do podcasts need transcripts?

Without one, deaf visitors get nothing from the recording, and people who cannot or will not play audio miss it too. Search engines also have almost no text to read.

What does WCAG 2.0 expect for audio?

For prerecorded audio such as a podcast, a text alternative presenting the same information, at Level A, its most basic level. In practice that means a full transcript.

Is a summary of the episode enough?

No. A summary or a list of topics leaves out most of what was said. The test is whether someone who cannot hear gets everything a listener gets.

Where should the transcript go?

On the same page as the player, as ordinary text, in full below it or behind a clearly labelled link beside it. Avoid images of text and careless PDFs.

Can automatic transcription do the job?

It gives a quick first draft, but it stumbles on names, accents and crosstalk. Someone who knows the subject should edit it before it goes on the page.

What about episodes already published?

Start with the recordings people still play, remove anything out of date, and work back through the rest. Every new recording gets its transcript before launch.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Recordings on your site with no text?

Send us the page addresses. We will tell you what each recording needs and the order to do them in.

Start a Conversation