Slashdot Mirror

I just wonder if ol' Staph has ever cybered with someone like himself on that DECtalk equipment... I was just playing with some online voice-synthesizers a week ago. Even downloaded some demos. Over at AT&T, I was having a good ol' time with a few "touch me in my no-no spots" and thare she 'blows. I'm so inclined to make a Stephen "cocci" Hawking rap for my voicemale.

http://www.nextup.com/TextAloud/index.html

I tested it, and besides the somewhat spartanic interface the quality of the voices (uses Festival?) leaves much to be desired. To start, it only supports a few languages; no German, no French, no Dutch. I especially need Dutch, English and German. It sounds absolutely robotic, even less good than Windows XP integrated voice. In contrast, check out the NeoSpeech samples. Especially the female voice at 16 khz sounds very clean. From what i've heard, the quality of the other voices is also just as good. These voices are of a much better quality than the solutions i tried on Linux (Festival) and only need a P2/400 while still allowing speed and pitch effect. Not sure if this is somehow usable on *NIX though.

The "incredibly realistic text-to-speech converter" isn't so much of a online desktop-app replacement but a demonstration of the AT&T Natural Voices extension for desktop software like ReadPlease and TextAloudMP3. Highly recommendable though.

When I was studying for my MCSE I took the compiled html version of my textbook, copied and pasted it into a speech synthesis program (text aloud mp3) and played it back at 200 words per minute.
It was remarkable how easy it was to digest the knowledge, even at that speed. I think that perhaps the synthetic voices allowed a bit more clairity than an actual human voice; as the synthisized voice does not use contractions like we're and you're (fairly Commander Data-esqe).
To augment the process I would read-along in my book with the voice and discovered that by stimulating more of my sensory input (and in my theory getting more regions of my brain active) I was able to plow through my books like a troop landing craft through a river.
perhaps this method of study, using both my eyes and ears (ocipital and temporal lobe) was so succesful because humans are supposed to learn, not just via one medium, but through as many sensory inputs as possible.
I remember hearing that smell can trigger very strong memories (makes sense since food is first smelt before consumed to verify it is healthy and unlikely to kill), perhaps by using scents along with lessons, learning can be further augmented.

Just start it up and copy the text you want to mp3 and it grabs it.

The best part-- is they have a program that does News from websites, just like you want it. I haven't tried that one though. I just wanted to listen to those boring manuals and exam crams instead of falling asleep reading them.

There is also Groups Aloud (for a News reader!!) and Stocks Aloud if you are still gambling in that arena. :-)

Well, I wouldn't be all that interested by this story except for that mention of time scaling.

I have been using my mp3 players of late (I have a d-link 32 meg thing and a Rio Volt) to listen to Project Gutenberg etexts converted to mp3 with TextAloud MP3 using AT&T Natural Voices speech engine. (You can hear a sample here.

Anyway, with time scaling, I'm guessing I'll be able to choose how fast to listen to my texts. That's exciting because your ability to listen to these things at high speed increases the more you listen. (As blind people can listen to audio books at surprisingly high speed, you will find your ability to listen increases as your practice increases.)

So, this seemingly minor feature could actually revolutionize the way you take in media.