4 ms·

I am surprised it’s taking so long for big tech companies to roll this out. I suppose there’s too much money in audiobooks? Perhaps licensing issues? If anyone

by lvl102 4y ago

I am surprised it’s taking so long for big tech companies to roll this out. I suppose there’s too much money in audiobooks? Perhaps licensing issues? If anyone can get a “celebrity reads aloud” feature that would be Apple and that could be big.

hxugufjfjf 4y ago
I cannot put into words how much I want Silmarillion read by Scarlett Johansson.
rickdeckard 4y ago
Yes, licensing. Narrated books are a separate publishing license, with its own production-cost model. So a tech.company like Apple already having rights to sell the written words is still not licensed to publish a narrated version. To sell an Audiobook, they have to acquire the license for a separate product, which so far includes the produced Audiobook itself. It seems Apple is trying to get the audiobook license directly from those authors who didn't sign the license away yet, undercutting production cost for the Audiobook with "digital narration" and then earning more money per sale... I guess we're going to see human narration die very fast now, at least for some common languages, and tech.companies want to ensure that they can split license from production cost, instead of being forced to buy the "whole" Audiobook...
- karmasimida 4y ago
  TBH it will depend on how those narrations turn out to be. Cherr-picking is easy, but I paid for this, it needs to be human quality throughout
  - rickdeckard 4y ago
    True, but I expect the critical mass of the target group will remain to be people frequently listening to Audiobooks --> Those people are more-likely subscribed to an Audiobook service --> are not paying a per-audiobook price but a flat-fee --> The flat-fee for the whole catalog is lower than buying one audiobook. I would expect this target-group to access a mix of human and digitally narrated books during the transition to digital narration, with best-selling books still being narrated by humans. Users may then complain about the quality of the digital narration, but will keep using such services as the price-expectation is now set. --> A competition for better digital narration engines will likely drive evolution of the engine and authoring tools, further increasing the pressure of publishers to justify the bottom-line of per-book Audiobook production costs. > Cherry-picking is easy, but I paid for this, it needs to be human quality throughout That's a really interesting aspect. If the Audiobook delivers the content with a human voice but still not engaging enough, how many listeners would put the blame on the narration rather than the book itself... ("I like this new song of Metallica, but I don't like how they sing it")
abudabi123 4y ago
I want on Spotify early faery tales and parables of dragon-slayers and dragons as read aloud by Joe Rogan, Michael Bisping, Tom Aspinal to kids of the lower political-economy.
kzrdude 4y ago
The quality needs to be very high for it not to be jarring. I stay with podcasts if the host has a "good radio voice". ("Not even all human voices are good enough for me.") It's just a very intimate medium to have someone's voice right in your head. If the voices have annoying quirks, those audiobooks will not be loved. Yes, it needs to be tens of hours of perfectly good narration.
- sidibe 4y ago
  I have used TTS for books for years. It might be jarring the first time but you get used to it very quickly, I don't even think about it. I actually prefer it to my Audible books usually because it doesn't ever do anything to annoy me like some narrators, and I can understand it at whatever speeds. There are some dialogue heavy books where I have to read along though to be sure of who is talking.
- lvl102 4y ago
  I think you’re underestimating state of the art in this area. You can do amazing things with just a few minutes of readings.
  - rockemsockem 4y ago
    No. You are vastly overestimating it. There is a reason there is no broadly available TTS service like there is for text-to-image. Anyone who says you can clone a voice in a few minutes is not talking about human-quality.
    - lvl102 4y ago
      I’ve done a few models on my own. Stephen Fry is a relative easy one. This is from 2020 so I am sure state of the art is far better now.
      - rockemsockem 4y ago
        I'm not saying you can't do it, I'm saying it likely does not sound good enough for the average person to listen to for a long time. Got a sample?

by lvl102 4y ago

hxugufjfjf 4y ago

I cannot put into words how much I want Silmarillion read by Scarlett Johansson.

rickdeckard 4y ago

Yes, licensing. Narrated books are a separate publishing license, with its own production-cost model. So a tech.company like Apple already having rights to sell the written words is still not licensed to publish a narrated version. To sell an Audiobook, they have to acquire the license for a separate product, which so far includes the produced Audiobook itself. It seems Apple is trying to get the audiobook license directly from those authors who didn't sign the license away yet, undercutting production cost for the Audiobook with "digital narration" and then earning more money per sale... I guess we're going to see human narration die very fast now, at least for some common languages, and tech.companies want to ensure that they can split license from production cost, instead of being forced to buy the "whole" Audiobook...

karmasimida 4y ago
TBH it will depend on how those narrations turn out to be. Cherr-picking is easy, but I paid for this, it needs to be human quality throughout
- rickdeckard 4y ago
  True, but I expect the critical mass of the target group will remain to be people frequently listening to Audiobooks --> Those people are more-likely subscribed to an Audiobook service --> are not paying a per-audiobook price but a flat-fee --> The flat-fee for the whole catalog is lower than buying one audiobook. I would expect this target-group to access a mix of human and digitally narrated books during the transition to digital narration, with best-selling books still being narrated by humans. Users may then complain about the quality of the digital narration, but will keep using such services as the price-expectation is now set. --> A competition for better digital narration engines will likely drive evolution of the engine and authoring tools, further increasing the pressure of publishers to justify the bottom-line of per-book Audiobook production costs. > Cherry-picking is easy, but I paid for this, it needs to be human quality throughout That's a really interesting aspect. If the Audiobook delivers the content with a human voice but still not engaging enough, how many listeners would put the blame on the narration rather than the book itself... ("I like this new song of Metallica, but I don't like how they sing it")

abudabi123 4y ago

I want on Spotify early faery tales and parables of dragon-slayers and dragons as read aloud by Joe Rogan, Michael Bisping, Tom Aspinal to kids of the lower political-economy.

kzrdude 4y ago

The quality needs to be very high for it not to be jarring. I stay with podcasts if the host has a "good radio voice". ("Not even all human voices are good enough for me.") It's just a very intimate medium to have someone's voice right in your head. If the voices have annoying quirks, those audiobooks will not be loved. Yes, it needs to be tens of hours of perfectly good narration.

sidibe 4y ago
I have used TTS for books for years. It might be jarring the first time but you get used to it very quickly, I don't even think about it. I actually prefer it to my Audible books usually because it doesn't ever do anything to annoy me like some narrators, and I can understand it at whatever speeds. There are some dialogue heavy books where I have to read along though to be sure of who is talking.
lvl102 4y ago
I think you’re underestimating state of the art in this area. You can do amazing things with just a few minutes of readings.
- rockemsockem 4y ago
  No. You are vastly overestimating it. There is a reason there is no broadly available TTS service like there is for text-to-image. Anyone who says you can clone a voice in a few minutes is not talking about human-quality.
  - lvl102 4y ago
    I’ve done a few models on my own. Stephen Fry is a relative easy one. This is from 2020 so I am sure state of the art is far better now.
    - rockemsockem 4y ago
      I'm not saying you can't do it, I'm saying it likely does not sound good enough for the average person to listen to for a long time. Got a sample?