Wednesday, February 11, 2009

Book publishers object to Kindle's text-to-voice feature

Book publishers object to Kindle's text-to-voice feature, article.

TTL pointed me to this one, and I echo his words about it: un-f***ing-believable!

I guess next thing we'll have Hollywood studios suing to get licence payments for camcorders sold because people could make their own movies and thus undermine Hollywood profits.

10 comments:

Alex said...

If Kindle had a "virtual-actor" not a "text-to-speech" utility I could see some merit to this.

Are there TTS utils which will look for how to speak? Do they try to do character voices, switching to a different voice for every speaking part in dialogues?

TTS reads very flatly, like a 6 year old reading a speech. This is not like an audio book. Audio books are a performance, with attempts to read with passion and emotion, not just straight vocalizing.

On a happier note I see Joss Whedon has released Dr Horrible as a region 0 DVD (despite phased release to iTunes), making the disk globally accessible, coupling with Amazon to provide burn on demand copies, so it's not limited by number of pressed DVD's.

Back to Kindle, I hope that the TTS feature does not get squashed. Maybe some DRM whereby if a talking book will be released then the TTS is disabled. However I find it a shame that the Harry Potter I could read for $15 costs a blind or illiterate person $50 for the CD narration. There's a one, can registered blind bibliophiles get reduced price audio books?

Eolake Stobblehouse said...

Prices on audio books are simply obscene.

Tommy said...

Gee, I was excited until I read "TTS reads very flatly, like a 6 year old reading a speech.", which of course makes total sence.

It would be a trick (nice if you could) to have the TTS read and interpret the text and then speak it correctly.

Alex said...

I was talking to a friend about how with CGI taking off, what if people could make a virtual actor. That is someone designs a physical being (mesh and textures I guess), and then the film company can use that one of these pre-defined characters, tweak weight, age and height to suit. It'd be like asking an actress to lose 5 lbs for a role, or asking someone to shave their head.

The next stage would be to have TTS voices which could inflect. Then an studio would buy in characters, voices, sets and just glue them all together with the plot.

I have read books where you get several quoted lines one after the other, and there are three characters talking. I end up losing track of who is who. I wonder how a machine would cope?

Do you think it would work best with plays or with books?

Eolake Stobblehouse said...

I'm sure it'll happen, but I'm unsure if it'll take 20 or 100 years.

Tommy, see this for voice sample:
http://tr.im/ftu1

Anonymous said...

The prices on audio books are high but that's why they invented illegal downloads. What saps actually pay for this stuff anymore?

Shaggyman said...

When you read text you rely on your imagination for nuance. There is no real reason why a flat voiced reader can't offer the same experience - different from an interpretive reading by an actor but if you have the same imagination you had when you read the paper edition you should still enjoy the book. I have been listening to computers read books while driving for twenty years, from the days when they were really hard to even understand. I have "read" about 150 or 200 books this way that I would never have read otherwise. The Kindle library is vastly larger than any audiobook company would ever, ever hire actors to perform, so for simple availability the Kindle has it all over the audiobook companies. That was why I turned to computer voiced readings in the first place, because of the larger selection -- mostly 19th century literature by the somewhat lesser known authors, no one was recording. I found the works as emotionally involving as their paper version. Not the same as a great reading by an actor, but since no actor was reading the things I wanted to hear, a perfectly acceptable alternative.

Eolake Stobblehouse said...

Thank you very much.

How did you get/make computer-read texts, especially portable, in the nineties?

Shaggyman said...

Back in the early nineties Soundblaster sound cards for PC came with a free copy of DEC-Talk, a gravel-voiced synthesizer named Dennis. It could read any text file in real time. The Gutenberg archive already existed with maybe a thousand .txt files of old books. In those days I would set Dennis to read and run a lead to a tape deck and make cassettes. I usually had to spend time fixing his dictionary to correct annoying mispronounciations.

Then came a program called iSpeak, a bit more intelligible, still strictly a synthsized voice. Same process though on the recordings, all cassettes.

Then came Text Aloud from Nextup.com. This program could save MP3 files, read at very high speed silently, and record a whole book in maybe half an hour. Then burn it to an MP3 CD. Plus the voices now were made by concatenative synthesis, in which a live actor would record word sounds matched to a huge dictionary. Each voice had some level of personality to it, and could respond sensibly to a question mark or a comma. Now I own more than a dozen voices. I prefer the British ones. I download the MP3 files to my iPod, carry it on my belt for walking the dog or yardwork, and have installed a car stereo with a cable in the glove compartment that runs the iPod with controls on the dashboard stereo.

The Kindle raises new possibilities but it is expensive and you would have to use a charger on the cigarette lighter and my stereo has a front jack I could plug it into to get the stereo amp and speakers. A bit more wiring than the iPod to use it, but a possibility. I am considering it but want to hear more samples of its voice first.

My main attraction to the computer voices was getting more literary texts than the audiobook companies were doing. The Kindle opens the door to a lot of nonfiction works that are very unlikely to ever be recorded.

The longest book I did with a computer was a 16th Century Chinese novel that must have been a thousand pages long in paper about a family of magic foxes who took human form and led a provincial revolt against the central government. It was quite fascinating, read by a British woman whose voice had been synthesized, quite a pleasant voice and very human sounding most of the time.

Eolake Stobblehouse said...

Wow, cool.

I only really got this idea after the mac voices got a lot better with Leopard.