Spotify Is Adding Auto-Generated Transcripts To Millions of Podcasts

Mia Sato writes via The Verge: Spotify is rolling out auto-generated podcast transcripts to more creators in the coming weeks, the company announced Thursday. The text transcripts will also be time-synced so listeners can visually follow along as a podcast episode progresses. Transcripts are available by scrolling down below the podcast player and tapping into a "read along" section. A transcription of a show makes the podcast more accessible to users and allows listeners to skip around and skim an episode without listening through.

Spotify says "millions" of podcast episodes will get the tool, and in the future, creators could add media to transcripts -- a useful feature if a creator is describing an image on the show, for example. Beyond transcripts, mobile podcast listeners globally will now be able to jump around an episode using chapters as well. Podcasters can add time-stamped chapters to their shows that briefly describe a segment of the show, allowing listeners to preview topics or start listening at specific points. The feature follows the recent addition of an AI-generated voice cloning tool that translates podcasts into different languages.

I have a brilliant idea!

[dskoll]

Just imagine if we transcribed the audio... and then did away with the audio! Then people could, ah, we need a verb... read at their own pace!

And maybe we could group many lines of transcript into things called... uh... pages. That's it!

And possibly group a large number of these pages into books that... uh... wouldn't even need power to read.

This will revolutionize society!

Re:I have a brilliant idea!

[VeryFluffyBunny]

Having a text in modes, i.e. spoken & written, can be very useful in some cases. The one that first springs to mind is 2nd & foreign language learning. The transcript can be used to consult new/unknown/unintelligible words & expressions. Also, reading & listening simultaneously, AKA "bimodal input" can be beneficial to language learners for various reasons.

Re:

[dbialac]

Of course. These things work great. I saw the description of a video on Amazon yesterday that talked about the September 7, 1941 attack on Pearl Harbor.

Re:

[jimbobxxx]

Just imagine if we transcribed the audio...

And possibly group a large number of these pages into books

You could put a collection of the book things into a big store. A bit like the way that software is put into libraries.

Re:

[tlhIngan]

Too bad you can't express emotions the same you in text as you do in voice. Things like enunciation, tone of voice and other attributes cannot be expressed in text so a joke that is hilarious in audio form can fall flat in a text transcript of same.

The reason for the need of transcripts is for ease of searching content.

You cannot take a series of interviews and then just transcribe them into text and expect to have a faithful reproduction of the original audio. I mean, if that worked, people would be sellin

Re:

[detritus.]

They won't do anything that brazen. What they'll do instead is automatically flag and remove/shadowban/depromote any problematic conversations transcribed and fed through their LLM AI bot. Same shit done for years on social media as well as youtube. Spotify is just catching up and applying the latest tech.

Re:

[Eunomion]

I've seen some pretty brazen shit from social media companies. Sooner or later, these companies are bound to claim that the very act of falsifying someone's uploads makes it their own original content. They'll hide behind "fair use" while denying everyone else its legitimate protection.

How good is it really?

[CoderFool]

My experiences with autogenerated captions are that it has a lot of trouble if the speaker mumbles or has a heavy accent. It also has trouble with jargon specific to a particular subject that is not part of the common language. Like diode or xamarin or c#.

Re:

[drainbramage]

I enabled closed captioning on my tv after seeing the auto-generated captioning at a friends house.
It was HILARIOUS!
The random delays and completely wrong word spaghetti reminded me of watching 60'S SCI-FI movies.
Much better than the actual content of the show.

Re:

[alishakihn]

I liked the idea