Computer read audiobooks - do it yourself

Everything except LibriVox (yes, this is where knitting gets discussed. Now includes non-LV Volunteers Wanted projects)
Post Reply
Hokuspokus
LibriVox Admin Team
Posts: 8015
Joined: October 24th, 2007, 12:17 pm
Location: Germany
Contact:

Post by Hokuspokus » August 29th, 2009, 10:28 am

Well, first off all, many, many apologies for this sacrilege. :oops:

But there are so many text out there we haven't recorded yet, and so many we can't record for many years, because of copyright.
I personally hardly find time to read books to see if I would like to record them and I always wished I could listen to text, before I record them. And today I learned how.

Many of you probably know that Adobe Acrobat Reader actually can read text loud, using the Windows computer voice Sam. I haven't tried that, so no idea how good or how bad it is. But I installed a German computer voice Steffi and I was surprised how good it is.
Not as good as any Librivox recording, of course, but quite understandable. (It can't say"nach" or "Buch", but says "ch" like in "durch", it says "Ziege - uner" instead of "Zigeuner").

You can get the computer voice Steffi for free here:
http://depositfiles.com/files/6212458

Now Acrobat Reader doesn't create mp3. But there are two free progis that can create audio from text.
Panopreter: www.panopreter.com
Easy to use, generates wav and mp3, doesn't fit into my netbook screen.

Balabolka: http://www.cross-plus-a.com/balabolka_de.htm
Generates wav, mp3 or ogg, looks like it can do quite a lot.

I'm quite happy with the result. And the good side effect is: I will enjoy LV recordings much more after listening to the computer voice for a while. ;-)

momof3Chihuahuas
Posts: 1304
Joined: June 9th, 2009, 7:11 pm
Location: Southern Indiana
Contact:

Post by momof3Chihuahuas » August 29th, 2009, 10:15 pm

I don't think you should feel bad, Hokuspokus; if it helps you listen to books, that's great! It's cool! :D And the technology is much needed, I think. I say this because I have a machine that reads books (I got it before I ever heard about Librivox). I have trouble reading a whole book all the way through because my eyes are extremely dry, plus they water, sting, & burn nearly all the time. Probably has to do with my inflammatory condition, ankylosing spondylitis (autoimmune illness that causes a lot of pain in lower back & other places; it can also affect the lungs & the eyes). I also have multiple allergies.

First, though, NO computer voice can ever compete with an actual person reading a book! There are problems with homophones (a computer voice can use the wrong pronounciation for a word that is spelled the same but has different pronounciations). And there are other issues.

I got SARA online from Freedom Scientific. SARA stands for Scanning And Reading Appliance. I've had it about a year. It's not cheap; I was able to get a reconditioned one for $1,000 off the retail price with part of my disability lump sum money (maybe I shouldn't say, but I paid $1,800! Yes, that's the discount price!). Still, it's worth it---every penny. I've listened to a lot of books on it, including The Time Traveler's Wife (which has recently been released as a movie in the U.S.). Basically, it's a large scanner. And it does take longer to read this way, because I have to scan in each page individually (I don't really mind); then I can choose from ten voices (male and female) to read the book. There are American voices (Tom, Jennifer, Jill, & Samantha), British voices (Daniel, Emily, & Serena), Australian voices (Lee, male, a bit too soft-spoken--& Karen; she's OK), & one Indian voice (her name is Sangeeta; I'd rather it be you, Neeru! :wink: ).

I've noticed that the male voices tend to sound a bit more robotic. I don't know why. My favorite voices, depending on what I'm listening to, are Samantha (U.S.) & Serena (U.K.). I could choose any number of people here who should've been the models for the computerized voices; it would take me too long to list them all! LOL :wink: The only other voice type I wish it had is Scottish (perhaps one male & one female).

I've listened to Serena read several Ellis Peters's Brother Cadfael books. :) I've gotten other Brother Cadfael books from Audible or itunes. I also get audiobooks from either of 2 physical (buildings) libraries & one that's online (Indiana Digital Media; but the selection isn't that great).

Oh, and SARA came with a headset that has a volume control on it---and she also came pre-loaded with 100 public domain books, & LV has all of them but two (The Red Thumb Mark & The Log of a Cowboy).

So, if I want to listen to a book that isn't in the public domain & the audiobook price is too high (or I can't get unabridged), I use SARA. Also, several months after I got her, I got a 19-inch flat-screen monitor on a pedestal; I have it hooked up to SARA & I can follow along with the text in ANY size font I want---I've got it set to about 30 or 34 right now, (& it can go much larger).

Sometime I'll put the voices on here if any of you want to hear them; I can record the various voices reading about a minute each from a public domain book. Poor Serena, though, can't even pronounce Heidi correctly! That's sad. It sounds like Hee-dee. LOL Oh, well.

Sorry to talk so long. All in all, though, SARA is a really nice machine & I'm glad I got her. Once I got used to the various voices (especially my 2 favorites), I could understand about 98 percent of what was/is being read without my monitor.

Technology can be a very cool thing! (Including the technology used to make LV books!) *big smile*
Joy :->
You cannot understand the glories of the universe without believing there is some Supreme Power behind it. Stephen Hawking

Twitter @momof3chis (but don't have cell phone LOL :roll:)

momof3Chihuahuas
Posts: 1304
Joined: June 9th, 2009, 7:11 pm
Location: Southern Indiana
Contact:

Post by momof3Chihuahuas » August 31st, 2009, 10:19 pm

Oh, dear, I seem to have killed the thread! Sorry about that, Hokuspokus. I think it's great that you have that cool technology.

If anyone else wants to reply to Hokuspokus, don't let my over-chattiness stop you, OK?? :?

Joy :)
You cannot understand the glories of the universe without believing there is some Supreme Power behind it. Stephen Hawking

Twitter @momof3chis (but don't have cell phone LOL :roll:)

Hokuspokus
LibriVox Admin Team
Posts: 8015
Joined: October 24th, 2007, 12:17 pm
Location: Germany
Contact:

Post by Hokuspokus » September 1st, 2009, 2:24 am

Never mind, Joy.
I think it is simply that people here are not that much interested in computer read audio books.
I wouldn't be either if we had so many audio books in German as we have in English.

annise
LibriVox Admin Team
Posts: 30000
Joined: April 3rd, 2008, 3:55 am
Location: Melbourne,Australia

Post by annise » September 1st, 2009, 2:35 am

I found it interesting and have started to look into it to see what they sound like - I have downloaded computer read books a while ago but found they drove me mad - but there are a few non PD bokks which I would like to listen to - you know the ones you buy because you are going to read someday but don't . And they are not the sort of books the library has on disk.

Anne

Hokuspokus
LibriVox Admin Team
Posts: 8015
Joined: October 24th, 2007, 12:17 pm
Location: Germany
Contact:

Post by Hokuspokus » September 1st, 2009, 2:59 am

Anne,
maybe you are interested in some details I've learned the last days.
There are 2 different types of voices for text to speech: SAPI 4 and SAPI 5.

There are many free voices for SAPI 4, but they sound very artificial (horrible). SAPI 5 sounds more natural, but there are no free voices around, except Sam from Microsoft (comes with XP) sounds very American, I don't understand it so well, even though I can understand our American readers very well, and the German voice Steffi, which is really good.

Balabolka uses both types, but SAPI 4 is no pleasure to listen to and SAPI 5 makes a nasty jingle sound with this software.
Panopreter only uses SAPI 5, no nasty jingle with this software, but no choice of voices either.

Sooner or later there will be more free voices available, which is very good for preservation and access to literatur.

But even with a perfect computer voice, it can't be compared with a human read LV recording. What we do here is a form of art.

annise
LibriVox Admin Team
Posts: 30000
Joined: April 3rd, 2008, 3:55 am
Location: Melbourne,Australia

Post by annise » September 1st, 2009, 3:21 am

I'll look into it . Most of our readers have fairly standard reading voices - with some words pronounced differently - if you listen to the accents collection you will see what I mean. But it must be hard with computer voices - I find it hard to get the inflexions right in some of the older english word patterns and I don't know how they would manage with words like bow.

Anne

redaer
Posts: 530
Joined: February 23rd, 2008, 1:31 pm

Post by redaer » May 1st, 2017, 10:41 am

Thanks for this topic, Hokuspokus.

Occasionally I need an mp3 audiobook which is not in the LV catalog. As this happens seldom I looked for a free software to produce it. On a Windows 10 system I use Balabolka with the Microsoft Speech Platform voices (e.g. Hedda for German).
I am quite content with the resulting audiobooks. Words are sometimes read wrongly. One can correct these with dictionaries (e.g. "vergißt" to correct "vergisst" which is taken at the end of the word for an abbreviation). As far as I know one can not use phonetic spelling in these dictionaries, therefore some foreign words are impossible to correct.

Any suggestions for free text to speech software which can produce mp3 files?
Experiences or recommendations how to use Balabolka? Links for Balabolka correction dictionaries?
redaer

Hokuspokus
LibriVox Admin Team
Posts: 8015
Joined: October 24th, 2007, 12:17 pm
Location: Germany
Contact:

Post by Hokuspokus » May 1st, 2017, 10:20 pm

Hi redaer,

I use Balabolka, too. Soon after this post 8 years ago I discovered settings that don't produce the nasty jingle I mentioned back then.

I never came across correction dictionaries and I don't think there will be any. You would need a different dictionary for every voice you use because different voices interpret the spelling differently. It seems that some of the pronunciation is built into the voice and there is no way to correct these rules.

I have heard about tts software and free voices for android that sounded quite good in the samples I heard, but as I have no android device, I can't tell you more about it. The name of the app was VoiceAloud.

redaer
Posts: 530
Joined: February 23rd, 2008, 1:31 pm

Post by redaer » May 2nd, 2017, 11:30 am

Hello Hokuspokus,

Thank you for the information.

I only use a Win System, so an android app will not help me. Perhaps there will be one day a free software where one can use phonetic spelling to improve the pronunciation of certain words.

About the correction dictionaries one can edit and export these. I thought heavy users of Balabolka may have inserted already a lot of corrections which others could use then. One would not have to write again hundreds or thousands of entries.

As example a few of my Balabolka German.dic entries:
Auferstehung*=Aufer-stehung
eleusisch*=e-leusisch
Häuschen=Häus-chen
staatsmännisch*=staats-männisch
suggestiv*=sugges-tiv
Vespucci*=Vesputschi
Wasserläufe*=Wasser-läufe

Kind regards
redaer

Post Reply