University of South Florida has an amazing program called Lit2Go, in which they have public domain books narrated by professional voice actors[0]. I have had mixed results with the quality of librivox recordings, though I very much respect and appreciate people who are willing to attempt it. Lit2go is wonderful quality.
> Since it is volunteer based, would it be obnoxious to start using some sort of rating system for the readers?
If it was done in the spirit of future improvement, perhaps by ranking them in various dimensions or being prompted with questions such as "How could the narrator improve?" rather than just "Feedback."
I would love this. With no disrespect to people volunteering their efforts, I was underwhelmed by the Librivox books I sampled. I'm extremely glad the project exists, though.
He's not on librivox, but the best audiobook narrator I've ever listened to is Ray Porter, who does a lot of books on Audible. His 14 by Peter Clines and the Fold by Peter Clines were two of my all time favorite audiobooks. Also his The Cartel and the Power of the Dog by Don Winslow were both great.
Another recommendation: Stephen Fry. He's also a famous British comedian. He has narrated all of the Harry Potter books and the first Hitch Hiker's Guide. Some of the best audiobook narration I've encountered.
(The second HHGTG is narrated by Martin Freeman, I think. It was part of the 2005 HHGTG movie release. )
If we're just going to recommend narrators: Simon Vance, who read all the Master and Commander books for Blackstone Audio is fantastic. Every character has their own voice and he can say mizzen-t'gallant-stunsl with out missing a beat
Another narrator recommendation: Michael Page. He's a Brit, good with character voices, good enunciation. I'm listening to his narration of Republic of Thieves by Scott Lynch and it's benefiting greatly from his narration.
Simon Prebble has narrated a lot of the books I have in my Audible library, notably The System of the World (all the Daniel Waterhouse bits) and Endurance.
I also liked William Dufris who narrated Cryptonomicon and parts of Anathem.
I find new narrators tend to be very grating the first time I start listening, then over time I get used to them.
On NPR, there was this Selected Shorts story read by William Hurt. I really liked it so I looked into the audiobook version. It was read by another actor and it was like night and day. I realized how good William Hurt was.
I was thinking about this the other day. Recordings have a potentially much higher quality than a speech engine. There a lot of reasons for this, but a couple of them, like mispronouncing words, or the wrong cadence, are fixable through human intervention in the engines interpretation of the text.
Wouldn't it be interesting if we had a speech engine that could take hints that were annotated in the text? Similar to how one can add type hints to clojure variables to clue the compiler in for better optimization. Perhaps it could even be unobtrusive and even readable, a la markdown, such as a tick mark to indicate a stressed syllable or earmuffs (asterisks) for an emphasized word. Perhaps inverted angle brackets >>could indicate speeding up a passage.<<
With such a system, it would be possible to use something like git/github to take "patches" for improving the speech of a piece of text.
This came to me when I was thinking about reading some of PG's essays to create audio versions that would be useful for long drives and accessibility. I emailed him asking for permission, but no response (I didn't expect one, he probably is flooded with email and/or didn't have the time for a response, but I thought I would try anyway.)
I suppose I could just start doing it, and if it garnered attention and ran afoul of YC's sensibilities, I'd get a response then.
However, it occurred to me that 1) my voice isn't particularly pleasant, and 2) audio files are very inefficient storage of textual information. Text to speech, while greatly improved over the decades, still isn't much of a substitute. But perhaps with hinting it could be good enough.
Interesting work! I've put together some thoughts on how markup/down/left might be used as slightly lower level tools to address some of these issues. It was great reading these because a lot of the problems simply hadn't come to mind when I was first thinking about this.
I want to dwell on it for a little bit (day or so) and then I'll send you an email.
You can see that an audio recording of a book published in 1921 was pulled due to legal letters.
Of course, in the US, works from 1921 are all supposed to be in the public domain. But already events like this occur for places other than LibriVox - multi-national companies with large legal departments and law firms on retainers not only assert rights for works from 1923 on, but for works before that as well. Most small companies and organizations can't deal with the time and expense required from these spurious legal threats and are forced to give in.
That would be a more valid argument if the link provided was about an actual copyright court case, rather than the estate of Agathy Christy asserting their ownership over the material in question.
Worth to note that Librivox also gives a significant boost to text-to-speech and speech recognition research, currently librispeech corpus based on librivox is the largest and quite useful corpus of spoken English. http://www.danielpovey.com/files/2015_icassp_librispeech.pdf
interesting, that book was one of my favourite when I first read it, but one of the parts I loved was the illustrations[0].
Just listening to the text should make up for such a different experience!
[0] I don't even know who did them in that edition, but they were in the style of gustave dore
I am an audio book fanatic, but I haven't been able to use Librivox very much. Quality of reading is everything and it is both a natural talent (voice) and an Art. An intelligent reading requires a great deal of preparation and a high level of dramatic intelligence. I would pay every time for a voice like Sean Barrett's.
Sadly the commercial market is totally dominated by Audible and the cut they take from publishers is enormous. So the price of great audiobooks will remain high. Also I can't listen to them on Linux (unless you count Android which I don't really).
What I'd love is:
* A new platform that is publisher friendly (perhaps an initiative funded by a group of publishers
* DRM free
I like to listen to podcasts and audiobooks to fall asleep. The android app[1] works well for me. Some of the Mark Twain[2] books were professional quality.
Some really good stuff to be found here. I recently listened to What I Believe by Leo Tolstoy, read by David Barnes (https://librivox.org/what-i-believe-by-leo-tolstoy), and both the reader and the recording was as good, if not better, than any audiobook I get from Audible. But it does vary. They also have a very nice iOS app.
[0] http://etc.usf.edu/lit2go/