Hacker News Reader

Comment by codegladiator | Hacker News Reader

> a distinction between a newly created piece of text composed using AI, based on an audio recording, and the potentially near-identical text version of the book the audiobook was created from

How is transcribing audio book "a newly created piece of text" ? This is another instance of a corp trying to use authors contents without actually paying for it (first being google AMP) in the name of experience.

Amazon's justification seems to be:

> Amazon says its transcriptions may contain errors and are not intended to be complete recereations of the text version of a book.

Really, what does this have to do with anything ? A thief claiming ownership because they didn't steal everything at once ?

View on HN

they Don't actually claim ownership. Audible has another product "Immersion reading" where you can buy both the Audio version and written version of a book and audible will sync between the two highlighting the spoken words on the written version in real time.The user can go forward/backward in the written version and have audio synced to that. Again this works with a licensed copy of both audio and text of the book. Whereas closed captioning is completely different. It is driven completely by the audio, does not have the accuracy of the real text, is being displayed line by line and in a time you don't control. noone would consider that equivalent to having the complete text book. And the intention to me is clear. Audible is already licensing Text versions for the customers who want them while having CC is a hearing aid just for the audio version.

Let's take that "newly created piece of text", and use a text-to-speech NN trained on the narrator's voice, and create alternative audio books. Maybe we can even include the original audio book in the training set for better results, after all we don't know that Amazon is not using the original book in theirs. I wonder if Amazon would consider it "newly created piece of audio".

If I buy the audio book does it make sense that I am allowed if I want to write on a piece of paper what I hear?

How about type it?

How about using some voice-to-text software that I have?

Yes, you do it for yourself. Thats fair use. Now you begin to sell it, thats an issue.

And to think amazon is "surprised and disappointed by this action" does not seem the case. They are definitely trying to make themselves look charitable because of "educational purpose".

So if they create a separate free app and inside it will allow to load books from Audible and transcribe them - you won't have any issue with it?

> My contract is crystal clear that the only rights conveyed to Audible are for voice recording and playback. The rights to reproduce text in any way are specifically withheld

The original creator has issues with it. And if I was the content creator, I probably would have.

The comment, under which we have this discussion, gives it very specific context: where will you draw the line in the range of tools to transform audio for fair use.

Why would you consider pen+pencil and free app to transcribe audio to text different in terms of copyright infringement? Does Audible charge you more to use this feature?

What I see here is an attempt to dictate ways in which I can use book I already purchased - i.e. limit my rights.

What I also see: a bunch of companies that are trying to squeeze more juice from the market, which they are not even willing to serve in the first place, because they, being experts in the book market, couldn't figure out that market needs this feature, while a bunch of IT geeks, did.

Will add just one last point on humane business practices: if publisher sold me the book so well that I decided to buy one of the most expensive forms of its distribution - he should feel lucky and fuck off.

The funny bit is that it’s been illegal to read books to your kids under similar reasoning from publishers.

It just was never enforced because how would that be possible.

But now with these ML/NN/AI tech approaches products are coming out at scale. If I write an open source package and give it away for free and it takes text and produces audio performances from it, that’s fair use, I think.

Amazon commercializing is a different thing, but unless the publishers are going to sue parents from using text to speech and speech to text utilities on stuff they bought we’re going to get into an interesting legal area.

Just like an MP3 is not an identical copy of music of the CD it was ripped from.

So, as long as the derived work is lower fidelity, everything is fine. /s

Reply on news.ycombinator.com