Slashdot Mirror
Creative Commons Audiobooks
xanderwilson writes "The New York Times (2nd half of the article; free reg. required as always) writes, 'Project Gutenberg is well known for offering free electronic versions of famous public-domain texts. Now Telltale Weekly wants to be its audio-book equivalent.' Of interest to others in the Slashdot community: Ogg Vorbis and MP3 downloads, payment via Bitpass micropayments, and a cheap-now, free later (with a Creative Commons License) business model." (And if you buy the Ogg Vorbis versions, part of the money goes to xiph.org.)
Five years or 100,000 paid downloads whichever comes first... yes I can support that model. Why the heck can't the RIAA or MPAA get with it??? nah, they've got to keep milking the cash cow for as long as they possible can... why else is stuff like Pink Floyd or Led Zepp's back catalogue so expensive still some thirty years after first release???
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
What a neat idea, I've been looking for some portable culture for my daily commute
You must not have looked very hard:
cat something.txt | festival --tts | lame - something.mp3
or something like that, I don't remember on top of my head.
I used to do that to get the news in my mp3 player automatically in the morning before hitting the road. Of course, it's not very convincing when it tells you something extremely sad or exciting, but it's understandable.
"A door is what a dog is perpetually on the wrong side of" - Ogden Nash
The idea is to slowly and continually fund, stock, and build a free audio library. Recordings of classic texts, which is the heart of Telltale Weekly, will be offered freely after five years or a given number of sales. When free, these audiobooks can be freely distributed whereever and however, including at Project Gutenberg, if they are interested.
Selling the work cheaply until then pays for current and future bandwidth, hosting, and recording costs--and attracts more talent to the project.
Alex.
Text-to-speech technology is no substitute for an audio book. Audiobooks are read by humans. Humans use slightly different voices for different characters, and infuse their voices with emotion. Some audiobooks are dramatized, with different readers for each character.
Would you take the script of a play or a movie, run it through tts and then say it was even a passable substitute for the original?
What I don't get is why they didn't choose Ogg Speex, a codec that is similarly Free, but aimed especially at voice recordings.
I know a thing or two about sound. I don't think that equipment is the big barrier - while audiophiles and sound engineers love to spend tens of thousands of dollars, the truth is that if you get a decent mic (about $100) for your PC, that will be plenty for spoken voice. Sure, it may not have perfect frequency reproduction and good noise rejection, but we're not recording a live band - just one person with no stray noise. You should also have a nice quiet room.
Where the cost comes into it is in the editing. Most people probably have acceptable voices - if you just teach yourself to speak at a good rate without stuttering. However, NOBODY, and I mean NOBODY, can read a page of text without any errors. Those nice audiobooks that you buy probably had 5 takes for every paragraph. If somebody misreads a sentence they probably just pause and reread it. Then the editor has to listen to the whole thing and splice out the errors. That takes TIME! Plus they probaby do multiple recordings of passages as necessary to get the right dramatic effect.
Then of course somebody has to "proofread" the final work for accuracy.
It is just like filiming movies - a nice digital camera is probably all you need to make a feature film, in theory (that and the sound equipment). However, the reality is that you need to film each scene from 14 angles 24 times and pick the very best clips for the show. That is what makes filming expensive.
I don't think that you'll ever see a completely free Gutenberg-like project for audiobooks - at least not until voice synthesizers sound just like people. Gutenberg works because of OCR and the ease of distributed proofreading.
Maybe the first step would be a distributed editing approach for audiobooks. If you could get somebody to do the initial reading, the editing could potentially be distributed. Granted, forget a simple web-browser interface - we'll need client-server at the least (potentially a Java applet might work), and lots of bandwidth. Still something worth thinking about though...
Ah, somebody understands....
Still something worth thinking about though...
At some point later this year I'd like to start a steering/planning discussion (forum or list, likely) about the direction Telltale will take to become more community-led. I'm fairly certain that by the end of the year, this project will be limited by what I'm doing with it, rather than encouraged by my work. If this is something that interests you, I hope you'll send me a note or join the newsletter.
Alex.