Would you benefit from audio versions of posts?

post by Raemon · 2018-07-26T04:53:20.733Z · LW · GW · 20 comments

I know two people who have a hard time reading, but totally enjoy audio versions of blogposts and were only able to consume the sequences via audiobook.

Curious how common this is, and how much low hanging fruit there is vis-a-vis either making audio versions of posts easily accessible, or finding a high quality text-to-speech thing to provide it by default.


Comments sorted by top scores.

comment by Dagon · 2018-07-26T16:56:37.116Z · LW(p) · GW(p)

I have a strong preference to read, especially for any posts with math or diagrams. I also seek out transcripts rather than watching video interviews or expositions. Audio would be zero-value for me.

Not negative, though; I don't object to making it available to those who require/prefer it.

Replies from: SaidAchmiz
comment by Said Achmiz (SaidAchmiz) · 2018-07-26T17:00:17.601Z · LW(p) · GW(p)


comment by ofer · 2018-07-27T06:25:40.528Z · LW(p) · GW(p)

It will be useful for me, I have lots of time spans in which I can only listen and not read.

comment by Ben Pace (Benito) · 2018-07-26T18:50:01.451Z · LW(p) · GW(p)

I think I would be excited to be able to listen to good blogposts as podcasts. Sometimes I just need to have something interesting to listen to while I play a mindless game and relax awhile.

Also absorbing important things through two different mediums helps me process - first via audio, to get an initial impression, and then via text to look closely at the details that felt wrong or important to me.

comment by Dan Scanlon (danscan) · 2018-07-26T18:13:01.381Z · LW(p) · GW(p)

I do a lot of traveling, and love consuming information via audio. I'm actually working on technology for natural narration of web content. I'd be happy to create high quality narration for all content on the site that you can embed into post pages. Feel free to reply or contact me if you're interested :)

Replies from: Raemon
comment by Raemon · 2018-07-26T19:33:22.750Z · LW(p) · GW(p)

Is the tech you're working on of the "quality text-to-speech" variety, or something different? Not quite sure I parsed your words.

Replies from: danscan
comment by Dan Scanlon (danscan) · 2018-07-27T14:57:42.824Z · LW(p) · GW(p)

I'm working on tech of the "quality text-to-speech" variety, especially for content that's longer than a few sentences, as a component of my project to make audio web browsing rich, delightful, and ubiquitous.

As an example, here's narration of this post [LW · GW]: (Narration of Fading Novelty).

Replies from: Raemon, Raemon
comment by Raemon · 2018-07-27T19:32:19.016Z · LW(p) · GW(p)

(For everyone else, curious if this quality would be sufficient to be beneficial? I'm not actually a podcast listener myself so I'm not sure what's considered good)

Replies from: Dr_Manhattan
comment by Dr_Manhattan · 2018-07-28T19:03:35.585Z · LW(p) · GW(p)

For me this does compare favorably to the TTS already built into the getpocket.com app, which also has a) follow along with text b) speed control features.

Replies from: danscan
comment by Dan Scanlon (danscan) · 2018-07-30T20:42:38.126Z · LW(p) · GW(p)

It'd be no problem to include speed control features in the first version, and text highlighting shortly thereafter (I already have this feature in my mobile app.)

comment by Raemon · 2018-07-27T19:31:35.008Z · LW(p) · GW(p)

Cool! Certainly that's among the better ones I've heard. (A couple hiccups here and there overall pretty solid)

note: still off the cuff and haven't checked in with other site devs, this is in the "curious how this might work" stage.

What's the process for that like? Right now is it just software running on your computer? How long does it take to render the sound? How resource intensive is it? Is there or could there be a webserver that our site could easily communicate with when new posts appear (or maybe when posts get over a certain karma, or moved to frontpage or curated)

Replies from: danscan, AABoyles
comment by Dan Scanlon (danscan) · 2018-07-27T21:40:12.190Z · LW(p) · GW(p)

Thanks! The audio was rendered via a narration web service I wrote for the backend of my project. It took 12 seconds to narrate and generate metadata (used to highlight the word or sentence that's being spoken).

I'd be happy to set something up such that you could post the contents of a new post to a webhook, and it could respond with embed code immediately. Placing the embed code e.g. under the post title would load a small widget you could play the audio from.

It's also possible to tune the narration quality specifically for LessWrong so it uses the proper lexicon and more natural emphasis :)

comment by AABoyles · 2018-10-19T14:39:10.144Z · LW(p) · GW(p)

Please please please make this happen!

comment by Dr_Manhattan · 2018-07-26T13:48:16.225Z · LW(p) · GW(p)

I occasionally use TTS in Pocket app (https://getpocket.com) (syncing the posts via their browser plugin). It's quite decent and I expect this technology to improve quite a bit in the coming years.

comment by mingyuan · 2018-07-26T16:14:00.430Z · LW(p) · GW(p)

It wouldn’t be useful to me, but might allow me to finally get my sister into rationality, since she’s basically incapable of reading :)

comment by Screwtape · 2018-07-26T15:22:36.523Z · LW(p) · GW(p)

Would not be especially useful to me, but my brother has a major preference to listen rather than read.

comment by ryan_b · 2018-07-26T14:24:32.819Z · LW(p) · GW(p)

It would not be useful to me.

comment by Elo · 2018-07-26T05:03:06.236Z · LW(p) · GW(p)

I use TTS. Fbreader app for Android, TTS plugin for that. Toepub.com to convert any file type to a .epub...

Completed RAZ this way.

comment by deluks917 · 2018-07-26T11:13:49.642Z · LW(p) · GW(p)

Would be useful to me

Replies from: AABoyles
comment by AABoyles · 2018-07-26T13:45:20.392Z · LW(p) · GW(p)

Me too!