Announcing: EA Forum Podcast – Audio narrations of EA Forum posts
We’ve started making audio narrations of some of the best posts from the EA Forum.
As of today, you can subscribe to the podcast:
EA Forum (Curated & Popular) Audio narrations from the Effective Altruism Forum, including curated posts and posts with 125+ karma. Subscribe: Apple Podcasts | Spotify | RSS |
EA Forum (All audio) Audio narrations from the Effective Altruism Forum, including curated posts, posts with 30+ karma, and other great writing. Subscribe: Apple Podcasts | Spotify | RSS |
Edit 2023-05-30: we reworked our podcast feeds a bit. We’ve updated the list above to reflect the latest selection.
You’ll also start to see narrations embedded on the EA Forum post pages themselves.
If a narration is available, you’ll see a blue loudspeaker button:
What can I listen to now?
Some of the winning entries from the EA Criticism and Red Teaming Contest:
Are you really in a race? The cautionary tales of Szilárd and Ellsberg by Haydn Belfield
Does economic growth meaningfully improve well-being? by Vadim Albinsky
Effective altruism in the garden of ends by Tyler Alterman
Notes on effective altruism by Michael Nielsen
Biological Anchors external review by Jennifer Lin
Some posts that were recently marked as “curated”:
Counterarguments to the basic AI risk case by Katja Grace
My take on What We Owe the Future by Eli Lifland
Population ethics without axiology: A framework by Lukas Gloor
AGI and lock-in by Lukas Finnveden, Jess Riedel, and Carl Shulman
What matters to shrimps? Factors affecting shrimp welfare in aquaculture by Lucas Lewit-Mendes and Aaron Boddy
Case for emergency response teams by Gavin and Jan Kulveit
What happens on the average day? by Rose Hadshar
How bad could a war get? by Stephen Clare and Rani Martin
What about AI narrations? I want to listen to everything!
The Nonlinear Library project is currently generating AI narrations of all posts that meet a fairly low karma threshold.
Within the next few months, we are hoping to collaborate with Nonlinear to develop a system that generates even better AI narrations for most or all EA Forum posts. We see a path to better pronunciation, emphasis, and tone, and also to much better handling of images, graphs and formulae.
Who is working on this?
This project is run by the EA Forum Team, in collaboration with TYPE III AUDIO.
What about the existing “EA Forum Podcast”?
In July 2021, Garrett Baker and David Reinstein started a volunteer project to narrate EA Forum posts. The most recent narration was published in January 2022. Since September they’ve been publishing Zoe and Coleman’s weekly summary episodes.
We are grateful to Garrett and David for their work on this.
We’ve not yet heard what they plan to do next—presumably they’ll release an update to subscribers in due course.
Thoughts, feedback, suggestions?
We’d love to hear from you! Please comment below, or write to team@type3.audio.
If you’re already listening to Nonlinear Library AI narrations, we’d be especially interested to hear what you think of them. What would you most like to see improved?
- Hiatus: EA and LW post summaries by 17 May 2023 17:17 UTC; 51 points) (
- List of AI safety newsletters and other resources by 1 May 2023 17:24 UTC; 49 points) (
- EA & LW Forums Weekly Summary (28th Nov − 4th Dec 22′) by 6 Dec 2022 9:38 UTC; 36 points) (
- EA & LW Forums Weekly Summary (5th Dec − 11th Dec 22′) by 13 Dec 2022 2:53 UTC; 27 points) (
- EA & LW Forums Weekly Summary (12th Dec − 18th Dec 22′) by 20 Dec 2022 9:49 UTC; 27 points) (
- EA & LW Forum Summaries—Holiday Edition (19th Dec − 8th Jan) by 9 Jan 2023 21:06 UTC; 24 points) (
- Best introductory overviews of AGI safety? by 13 Dec 2022 19:04 UTC; 21 points) (
- Best introductory overviews of AGI safety? by 13 Dec 2022 19:01 UTC; 21 points) (LessWrong;
- EA & LW Forum Summaries (9th Jan to 15th Jan 23′) by 18 Jan 2023 7:29 UTC; 17 points) (LessWrong;
- EA & LW Forum Summaries (9th Jan to 15th Jan 23′) by 18 Jan 2023 7:29 UTC; 14 points) (
- Hiatus: EA and LW post summaries by 17 May 2023 17:17 UTC; 14 points) (LessWrong;
- EA & LW Forum Summaries—Holiday Edition (19th Dec − 8th Jan) by 9 Jan 2023 21:06 UTC; 11 points) (LessWrong;
- EA & LW Forums Weekly Summary (28th Nov − 4th Dec 22′) by 6 Dec 2022 9:38 UTC; 10 points) (LessWrong;
- EA & LW Forums Weekly Summary (12th Dec − 18th Dec 22′) by 20 Dec 2022 9:49 UTC; 10 points) (LessWrong;
- EA & LW Forums Weekly Summary (5th Dec − 11th Dec 22′) by 13 Dec 2022 2:53 UTC; 7 points) (LessWrong;
- 31 Dec 2022 16:38 UTC; 5 points) 's comment on Future Matters #6: FTX collapse, value lock-in, and counterarguments to AI x-risk by (
- 7 Dec 2022 18:31 UTC; 4 points) 's comment on Winners of the EA Criticism and Red Teaming Contest by (
- 16 Apr 2024 3:22 UTC; 3 points) 's comment on A trilogy on anti-philanthropic misdirection by (
Could I record the audio of my own post?
Joe Carlsmith started doing this for some of his own blog posts, and Holden has narrated the posts in his ‘Most Important Century’ series. Maybe these author-recorded narrations could be incorporated into the EA Forum posts and/or podcast feed?
Good idea, thanks. I’ve shared to the EA Forum Team.
Irena: good question!
Personally I’m keen on the idea of post authors narrating their own writing. But I need to check with the EA Forum Team before I share a proper reply. I’ve sent them a note, and I’ll post back here on Friday if not before.
In the meantime: TYPE III AUDIO has some advice on how to make a great narration: https://blog.type3.audio/how-to-record-a-great-narration/
This is a very nice development. I’m a big fan of Radio Bostrom and am excited to see the same team of narrators record audio version of EA Forum posts.
One suggestion: when narrating linkposts, such as AGI and lock-in, it would be great if you could narrate the webpage/document to which the link points, rather than the post itself.
Thanks for the kind words!
Thanks also for the suggestion re: narrating the full document, rather than the summary. This does seem like it would be valuable in many cases and feedback like this is likely to push us in that direction :)
Co-founder of EA Forum podcast here. I wanted to say that I think this is great. As I’ve discussed in conversations with some of the OPs:
I just want there to be good audio content, which is why we started that podcast. I’m super happy that someone else is doing it, with professional skills and tools and that there are dedicated resources devoted to it. And it seems to be going very well!
As for the “EA Forum Podcast”
I’d be happy to merge all of that content with this new initiative, ideally including the embedings (note, we generally put the ‘audio’ tag on posts we narrated)
acknowledging that our audio and narration quality is general a bit lower; but hopefully still useful - and I sometimes gave comments in my own narrations …
Would be happy to semi-close that podcast and have it point here to avoid clutter and confusion, but Garret may have other ideas
Some suggestions for this podcast … possibly something we could top-up on “EA Forum Podcast”
Comment threads: I’d really like to see at least some of these narrated. Much of the value for many posts is even greater in the threads then in the original post. And of course it’s valuable to hear what people think about a post’s ideas and credibility, and get these other perspectives.
Author narration: I think this is often high value because it gives good insight into the tone and intentions of the writer. Of course this trades off with non-professional narration and not all authors having the audio tools. Still, it seems like something that I’d like to see happen more.
...
Congrats on a great launch. Really enjoying it and getting a lot out of it.
Great to hear, thanks David!
I share your enthusiasm for author narration, but I think it only makes sense when:
Some people love the idea of narrating their own writing, to the point where I slightly worry that some will fail to realise that (b) is true of them. But this is probably overthinking it.
My general take is: if an author is keen to narrate, we’ll briefly assess (a) and (b), and if there’s not an obvious issue there, we’ll gladly help them make a great narration.
The steps you suggested for the EA Forum Podcast sound good to me. I’ll send you an email later today.
Again—thanks for all the work you’ve put into the “OG” EA Forum Podcast B)
Narrating the top comments for some posts is a good idea. I’ll discuss this with the team.
My quick personal take is that we should start experimenting with this immediately. We should also bear in mind that not all posts will have comments that are above the bar for narration.
I actually suggested we narrate a couple of the replies to the Michael Nielsen post, but I lost track of that thread and I can see that this didn’t happen. I still think we should do this—I’ll discuss with the team.
Where can I find/filter the human narrations only?
At the moment I don’t think you can find human only narrations—but now I’m curious why you ask? (to be clear, we are evaluating how users feel about human vs AI and about a combined feed, so opinions are welcome)
If I can at least see which are which that would help.
Our plan is to move to 100% AI soon, unless we get feedback that suggests users much prefer human narration.
(we may have some exceptions, like the best classic posts or the EA handbook)
I was fearing that.
How will you know whether the preference is strong enough to justify human narration?
To be frank I don’t have a clear threshold in mind, but some points we have in mind are:
We can narrate all posts on the Forum with AI for the same price as a few posts a week with professional human narrators
Most engagement on posts happens in the first few days of publishing (definitely within the first week), which makes coordination to publish a human narration tricky
We expect AI narration to get better and better—already we think Type 3 Audio’s service is better than the NonLinear service (my subjective opinion, and largely because Type 3 is just using newer APIs)
All these factors make the threshold pretty high to keep paying for human narration. That said, we can definitely support people who narrate their own posts to override AI narration.
Good points. Some counter-points:
I assume that AI narration will converge to being close-to-costless at some point, so definitely that should be available. But these are not mutually exclusive—I think we should have some human narration; it need not be professional. People seem willing to record some stuff for free, and often do a decent job (or at least a job that I personally prefer to the AI, not all will though.)
But A. I’d rather us encourage longer more extended discussions and returning to the issues, rather than feel a rush to respond right away. Most posts are not time sensitive.
B. People get value from listening even if they don’t respond in a comment
True. I assume it will get so good (and apps to enable text to speech) at some point that people do it all themselves, and curate it all the way they like. At that point the ‘human readings’ might still have a use, though, for people who really like the human touch (but I’m not sure).
If cost of human narration is crowding out ‘much more AI narration’ than I would agree. But as AI narration becomes fairly costless than maybe it’s not a tradeoff.
I’d also suggest maybe supporting (i.e., hosting) narrators who do a decent job of narrating others’ posts as well.
The ai voice is growing on me. I like that it announces indented lists etc. It’s nearly as comprehensible as a human, with only occasional glitches.
But I still find it less pleasant. Miss the human touch.
I think there should be a platform where humans can narrate this stuff.
I think that people like doing it (people even volunteer) and people like listening to it. So it’s win-win to keep such a platform open.
I find myself far more interested in and attracted to the human narrations.
If I’m super interested in the post I’ll also listen to the machine one but it requires more focus.
Also, some machine narrators might be better than the voice you are using? Maybe that Obama-like one?
I still find this very useful, but I think the automation here is lagging a bit behind the technology in terms of realistic voices and pronunciation errors.
I’ve been hearing a lot better AI-generated narrations of books, news, etc., using the newest models. (I was even listening to an audiobook and didn’t realize it was AI for about 15 minutes until it made a glitch).
Any chance this could be upgraded?
Thanks for your feedback.
For now, we think our current voice model (provided by Azure) is the best available option all things considered. There are important considerations in addition to human-like delivery (e.g. cost, speed, reliability, fine-grained control).
I’m quite surprised that an overall-much-better option hasn’t emerged before now. My guess is that something will show up later in 2024. When it does, we will migrate.
Happy to see this, nice work!
Nice! Can you provide an update when it’s ready on Google Podcasts?
Google Podcast links:
All audio
Curated posts (coming soon… I’ve emailed Google support team)
Summaries
I’ve added these links to the OP.
Sure, I’ll post back.
I submitted a week ago and they’re taking ages. It’s known that they sometimes take a while but this case seems unusual. Dashboard shows no errors but if it’s not there by Friday I will try to find a Googler to speak to...
[This was a spam comment. The Forum support team has removed it.]
You wrote “while I wait for your response”. What are you referring to? I’m not aware of any comms with you aside from this comment.