Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Top HN articles in human-read podcast format
42 points by ksaitor on June 14, 2017 | hide | past | favorite | 40 comments
Hi HN,

Like many of you, I found myself discovering dozens of great posts on HN and never actually finding the time read them. I used to save every cool link to Pocket, but… well… same story — I rarely found the time to read them.

So I thought it'd be great to listen to the top articles, the same way I listen to podcasts. Initially, I tried out narrating articles with latest text-to-speech synthesis from Amazon and Google, but it was still pretty bad. Especially with long form content. So I thought I'll do this with real humans, real voice actors. So I made ReadByHumans. [1]

We are starting out by narrating top articles from HN and some longer cryptocurrency white-papers. Giving away 3 top articles from last week and we’ll be sending one more audio article weekly.

In future and at scale, we’d love to narrate any article or a document, on demand and with only a few hours turnaround time.

Would love your feedback! What type of content would you like us to narrate? Is it easy to access the podcast feed?

Thanks!

[1] https://readbyhumans.com




Have you thought about the rights situation here? I don't see anything that indicates that you secured the copyright of the articles you've already provided, and the sites themselves don't have any copyleft notices. I suspect that unless you have ongoing relationships with authors, it's unlikely that you'll be able to both respect their copyrights and turn around "any article or document" in "a few hours turnaround time".


I'm interested to see what the legal ramifications would be here. I don't see anybody objecting to a screenreader reading the content they have created. So what's the difference between that and a better-inflected human voice? Even with the screen reader, people have probably essentially paid to have copyrighted content read aloud.

So what's the difference?


Here's one huge difference: https://news.ycombinator.com/item?id=14550931

(Specifically, this site is depriving me of income.)


Thanks for mentioning screenreaders, Andrew! Very good point.

My initial idea was a browser extension, akin to Pocket, where you click on any article, and then get a narrated version into an app or your personal podcast feed within a few hours. I was afraid it would be very expensive to execute if ton of people would request a vast range of articles, and recording would get little reuse. So I limited articles selection to HN only and decided to launch like this and test the demand.

I agree with the other commenters here, and starting to see how distribution makes all the difference. Again, my intention was not to distribute or to re-distribute the content. But just to help consume it, when there is no time to read.

I'm considering taking the following steps: - turning all possible ways accessing narrated versions from the site — so we are not distributing it. - linking up to Pocket API or building a separate Chrome extension, where users can submit what articles they'd like to listen to. And then providing them this on a personal basis. It'll be a "human screenreader".

What do you guys think? Would you like to use this kind of "human screenreader"? Would content creators object? What are the copyright implications?


With a screen reader you have first to go to the original content site. You may have to log in, you will probably see some ads. All of this gives proper attribution and money to the authors pocket. This kind of service goes all around that. The fact that they did not even ask authors first is mind boggling.


So what's the difference?

Distribution.

Screen readers apply their transform to source content. By becoming that source (regardless of transform) you are distributing it with control of all the benefits involved, such as community, traffic, and ultimately revenue.


The difference is that screenreader does not aggregate multiple content sources behind its own wall with the intention of redistributing through its own channels (possibly for money).

Imagine Google AMP or Facebook articles with no way to opt-out (and maybe subscription fee).


A screen reader acts on behalf of an individual user. This is rebroadcasting someone else's content.


Seems like everyone here, even me, has had this idea -- but to my knowledge you are the only one to do it, so props for that!

The person doing the reading has kind of a funny way of reading it no? Bit of an accent, odd emphasis, lack of emotion, robotic style.

My opinion is that I won't use it with this narrator, but with a better, more practiced english orator I would.


This is some actionable feedback, @dharness! Thank you. That's one of the challenges of executing this — is finding good voices that a lot of people love listening to. We are working on this! :)

Btw, try listning to this voice on 1.2, 1.5, 2x speed. That's how i typically listen to my podcasts. Sounds perfect to me.

Also, what do you think about our female voice? https://readbyhumans.com/convert-case-studies-into-podcasts There is a demo there. Would you prefer that voice, over our male voice?


I suspect many of us who had this idea figured it would be copyright infringement, and solving that issue, especially in a morally non-ambiguous way would be difficult. At least that was it for me.


Likewise. I've thought about doing this for years, especially since there are so many great works that are rarely read. Now I'm afraid that it'll be haunted by the specter of the time when someone did it and infringed my commercial work and I yelled at them for it.


Oh this would be an amazing feature. The ability fine tune the human reader. Saying something like, I prefer female voices but instead I'd like someone with a different tone.


Had the same idea some time ago but did not have the same level of motivation that you did :D. Lots of kudos. One question I wondered about was the copyright perspective especially in on demand readings. An idea I had was to sell the service to publications. So maybe that's something you'd like to try in the future :).


Thanks, nstart! Well, just validating an idea, and trying to find a niche with ton of demand.

I did try out to sell this as a service to publications a few weeks back, but it seemed like a sales cycle would be way too long and that was not exactly solving my own problem, so I was not as motivated to continues talking to publications. Maybe I'll get back to that later.

I talked to 3 lawyers so far. There are ways to making it work. One way it to set up revenue share with all parties, which is ton of effort, ofc. Another one it to sell the service as a personal narration service, and not as a audio content itself. It seems to be a thin line, and I'm still diving into the details of that. One way or another there seems to be a way to make everyone happy and avoid legal issues. In my opinion, it's mainly a matter of demand. If the demand is high enough for such service, and opportunity is big, then it'll justify efforts, no matter how hard, to make this work.


On Product Hunt [1] Chris Messina mentioned a few ideas how to make legal part of this work

[1] https://www.producthunt.com/posts/readbyhumans


Some have said this is copyright infringement. Gary claimed so as well. Keep in mind, AFAIK, he is not a lawyer. (Neither am I) but ask your own! There are fair use laws, and especially if you created a trans-formative work, this could be legal! Good luck.

I'm personally un-sympathetic to Gary's plight to keep his work super secret. This feeling started when I found out that he asks conference organizers to NOT share the recordings of talks he gives. I'm not sure how often this has happened, but it feels against the spirit of presenting against these events.

If I've gotten something wrong, I'm sure someone will correct me, but those are the facts as I see them today.


When I accept an invitation to speak at a conference, I say "Yes, but I don't want to you to publish a video. If that's OK, then I'm in." Also, you may not realize this, but a lot of conferences are for-profit. It's not as if I'm tricking them by springing that constraint on them after the fact, and many of them are making money from the speakers' free labor to begin with.

Second, many many people have tried to build screencasting businesses. One of them turned into an 800 lb gorilla. Most of the rest don't exist any more. AFAIK, DAS was the second subscription-based screencasting company to exist, with the only prior art being a short-lived subscription option offered by TekPub (which was later acquired by the 800 lb gorilla). DAS still exists today, is still independent, and is still (almost) my only source of income, while dozens or maybe hundreds of others have come and gone. I'm sorry if you don't like the way I exercise control over my wholly-owned works, but my business model combined with control over my content has allowed me to do this successfully for six years (modulo a very long break in the middle). And I've never charged money for anything recorded at a conference; that wouldn't feel right.


Thanks for your feedback, James!

Check out my reply to andrewstuart2's post above. Seems like there is a way to make this work. Really excited and appreciative of all the comments here.


I started doing this a while back, too! However, I noticed recording takes way more time than I thought it would, and I didn't think I'd be able to keep up with the workload every day. The plan was always to involve the community and different readers, but I'd have to get it started: build a contributors website, do quality control, read stuff myself, etc. It never happened.

Would it help if I contribute a few articles a week? My twitter handle is the same as my username, or see my profile for email.


Thanks, Luc! Definitely, would love to collaborate! Will shoot you an email in a few min. Thanks for your support.


If you like these audio article concepts, check out SpokenLayer. They work with many big publishers.

http://www.spokenlayer.com


Thanks, @reustle! I discovered them some time ago. They seem to be doing a good job. Still a bit different from what I'm trying to do. The problem that I'm trying to solve is: tons of written content that I'd like to read, yet I don't have enough time to read it. So basically I'm trying to make a human screenreader. And trying to find the best way to make it happen :)

Thanks for sharing spokenlayer.com again. Good to know people are familiar with them!


o_O really? I must be living in the wrong universe. I often wish there were less tutorial videos, podcasts and such and just give me the text. I read much better than listen. Must be me. There's also of course the language barrier although I did get a 8 out of 9 on my IELTS for understanding spoken English and I live in British Columbia for some years now -- but then again, reading comprehension was 8.5 and to this day I have no clue where I lost that half point.


Nice work, Raman. I have one question and 1 feature feedback.

How would you scale this up when there are more demands if the product needs real human to read every article? Maybe that can be your business model (commercial reading for private podcasts) - there will be copyright issue though

One feature I think useful for me is that it should be searchable (by voice command is best) and read it from there instead of listening the whole article.


Scaling is doable. We are relying on real human speech at that moment.

Search — that's an interesting idea. That will probably come in a bit later, when there is tons of demand and we'll be able to pour efforts in all the latest AI q'n'a tech :D


Had a similar idea once upon a time to auto narrate any article using text-to-speech. Got about as far as you did it sounds like. Having human narration seems so plausible if you approached gig style. Between-work actors could read for you instead of driving for Lyft.


Exactly! How long do you think it'll take for TTS AI to catch up and be indistinguishable from human tone and voice?

Yeah, would be great to turn this into "next million jobs", where anyone can go and narrate content.

I'm still figuring out what community to target. Where people would love to listen to that kind of content. What do you think?


Our intended market was business/enterprise users wanting listen to briefs, articles, etc. like on their morning commute.


If anyone wants to do something similar to this in the command line (simple article text to audio) I have a little one line shell script here. Put the text inside article.txt and then run it. Mac only.

https://gist.github.com/reustle/7029e02e171f8a4b13d0


So, is it actually "by humans" or it's a synthesized speech? Just wondering, sounds a bit robotic :)


Humans. Perhaps the accent makes it sound robotic :) If you listen to the full article versions in your podcast app, you'll notice the difference.

Several people asked this question so far. It's fascinating that people doubt and not sure whether it's AI or person :D


Great idea! And I like that it comes straight to my email. Can multitask and listen at 2x speed!


Thanks, shnraj! What type of content would you like to be narrated?


Interesting idea. What articles did you narrated so far?


Thanks, Andrey! We've narrated a few top articles from last week:

- Options vs. cash. By Dan Luu link — [1][2]

- College ROI. By Erik Rood link - [3][4]

- Network protocols. For programmers who know at least one programming language. by Gary Bernhardt - [5][6]

And the original Bitcoin whitepaper by Satoshi Nakamoto. https://readbyhumans.com/doc/bitcoin-a-peer-to-peer-electron...

[1] https://danluu.com/startup-options/

[2] https://readbyhumans.com/doc/options-vs-cash-40ncs6z7q

[3] http://erikrood.com/Posts/college_roi_.html

[4] https://readbyhumans.com/doc/college-roi-9jyu8a764

[5] https://www.destroyallsoftware.com/compendium/network-protoc...

[6] https://readbyhumans.com/doc/network-protocols-for-programme...


I posted a copyrighted work on Destroy All Software, which offers paid subscriptions to that same work. You found a link designed to be shared as a sample and decided to copy my commercial work in full, on your website, behind your own login wall. You didn't ask first, or even tell me that you'd done it.

Links to that work convert into paying customers, which is how I buy food and pay my mortgage. You're not only infringing on my copyrighted work, but also depriving my small business of revenue. Remove the infringing copy of my work from your site immediately.


Hi, Gary. Content removed, as requested. Appreciate your patience!

Please understand me. No harm intended.

I just launched this to test an idea. And a lot of people seem to have a similar demand for this type of narration service. I'm not exactly sure how to make this happen, but i'm just testing an idea.

If you were building a service that solved this problem (way too much written content, not time to read, so let's listen instead) — how would you go about it?


If I were building this service, I'd do what several people in this thread have told you to do, and what I've already told you to do on Twitter: I'd ask for permission and only record an article after that permission is received.

As I just said to you on Twitter (https://twitter.com/garybernhardt/status/874884772174196737), this is blatant and willful copyright infringement. Maybe the owners of the other works that you're currently infringing on don't care. But if you continue to do this, you will eventually encounter someone who does care. And if it goes on for a while, and then they notice, and then they sue you for lost revenue (which you will probably settle extremely unfavorably because the infringement is so blatant) then it will be neither enjoyable nor profitable for you. The American legal system doesn't care whether "no harm [was] intended" if you are depriving people of the use of their copyrighted works.

Pull all infringing works from the site temporarily (which means every recording). Email the authors of the infringed works, asking them for permission. Never post another recording without permission. Maybe it will be an awesome service.


Oh nice! Thanks for sharing! Especially for Bitcoin article!




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: