Slashdot Mirror

Four years ago, my company was playing around with setting up some basic macros for L&H Systems (now ScanSoft and previously Dragon Naturally Speaking) to take separate USB Voice Note entries and formatting them for use with Contact Management, Word templates, etc. Why not get a high-quality USB-enabled voice recorder? Simply save the files separately, and run it through some decent Speech-recognition. Heck, even the training sessions can be done on the road and analyzed 'offline'. Still geeky, since you get to create some neat macros to BLOG those entries automatically. Maybe you can even use some logic to coordinate the rough times/dates with your digital pictures and automatically post them in chronological order with the pictures. Hmmm...Automated digital publishing...yummmy...

think I remember a demo of L&H's text to speech, and it wasn't much of an improvement over the ancient typical monotone voice. I don't remember if it was them or someone else. Does anybody have info on them?

I do, as a matter of fact, because I work there ;-)

The voice in Voice Xpress is indeed the typical monotone voice. But the latest and greatest voice synthesis technology is our RealSpeak. A demo can be found here. You can type in anything you want (up to 30 english words), and RealSpeak sends you a .wav file. I think you will agree that it sounds much better than the older computer voices (who sounded like a guy with a cold talking in a tin can).

To actually get people to use this on a regular basis, I think they're going to need a smoother, more natural sounding text to speech engine than what is currently available. I wonder if the technology is ready for this...

You might want to look into Lernout & Hauspie's new text-to-speech engine, RealSpeak.

First, listen to the pre-recorded samples (in several languages!); then use the Web demo to plug in your own text (I recommend a random article off a news site such as CNN). Ignore the "30-word limit"; it's bogus. For best results, listen to it read an article that you haven't read, and don't read along. I think you'll be amazed by the quality.

The voice is not quite convincing as belonging to an actual human being, but I still think it's a monumental achievement, especially considering the level of expressiveness it achieves with plain-text English (no hard-coded phonemes or stress codes). What's more, it's by far the most comprehensible and pleasant-sounding TTS engine I've ever encountered -- and, being a TTS nut, I've played with a lot of them, as far back as "Speech by Andy Maguire" on the IBM PC internal speaker and even S.A.M. on the Apple ][.

No, I don't work for L&H, but I find their latest TTS engine exciting, and I plan to snap up a copy (as well as the SDK) as soon as it's released in the form of a mainstream commercial product. I can only hope that Ananova sounds as good as RealSpeak.

To actually get people to use this on a regular basis, I think they're going to need a smoother, more natural sounding text to speech engine than what is currently available. I wonder if the technology is ready for this...

You might want to look into Lernout & Hauspie's new text-to-speech engine, RealSpeak.

First, listen to the pre-recorded samples (in several languages!); then use the Web demo to plug in your own text (I recommend a random article off a news site such as CNN). Ignore the "30-word limit"; it's bogus. For best results, listen to it read an article that you haven't read, and don't read along. I think you'll be amazed by the quality.

The voice is not quite convincing as belonging to an actual human being, but I still think it's a monumental achievement, especially considering the level of expressiveness it achieves with plain-text English (no hard-coded phonemes or stress codes). What's more, it's by far the most comprehensible and pleasant-sounding TTS engine I've ever encountered -- and, being a TTS nut, I've played with a lot of them, as far back as "Speech by Andy Maguire" on the IBM PC internal speaker and even S.A.M. on the Apple ][.

No, I don't work for L&H, but I find their latest TTS engine exciting, and I plan to snap up a copy (as well as the SDK) as soon as it's released in the form of a mainstream commercial product. I can only hope that Ananova sounds as good as RealSpeak.

To actually get people to use this on a regular basis, I think they're going to need a smoother, more natural sounding text to speech engine than what is currently available. I wonder if the technology is ready for this...

You might want to look into Lernout & Hauspie's new text-to-speech engine, RealSpeak.

First, listen to the pre-recorded samples (in several languages!); then use the Web demo to plug in your own text (I recommend a random article off a news site such as CNN). Ignore the "30-word limit"; it's bogus. For best results, listen to it read an article that you haven't read, and don't read along. I think you'll be amazed by the quality.

The voice is not quite convincing as belonging to an actual human being, but I still think it's a monumental achievement, especially considering the level of expressiveness it achieves with plain-text English (no hard-coded phonemes or stress codes). What's more, it's by far the most comprehensible and pleasant-sounding TTS engine I've ever encountered -- and, being a TTS nut, I've played with a lot of them, as far back as "Speech by Andy Maguire" on the IBM PC internal speaker and even S.A.M. on the Apple ][.

No, I don't work for L&H, but I find their latest TTS engine exciting, and I plan to snap up a copy (as well as the SDK) as soon as it's released in the form of a mainstream commercial product. I can only hope that Ananova sounds as good as RealSpeak.

To actually get people to use this on a regular basis, I think they're going to need a smoother, more natural sounding text to speech engine than what is currently available. I wonder if the technology is ready for this...

You might want to look into Lernout & Hauspie's new text-to-speech engine, RealSpeak.

First, listen to the pre-recorded samples (in several languages!); then use the Web demo to plug in your own text (I recommend a random article off a news site such as CNN). Ignore the "30-word limit"; it's bogus. For best results, listen to it read an article that you haven't read, and don't read along. I think you'll be amazed by the quality.

The voice is not quite convincing as belonging to an actual human being, but I still think it's a monumental achievement, especially considering the level of expressiveness it achieves with plain-text English (no hard-coded phonemes or stress codes). What's more, it's by far the most comprehensible and pleasant-sounding TTS engine I've ever encountered -- and, being a TTS nut, I've played with a lot of them, as far back as "Speech by Andy Maguire" on the IBM PC internal speaker and even S.A.M. on the Apple ][.

No, I don't work for L&H, but I find their latest TTS engine exciting, and I plan to snap up a copy (as well as the SDK) as soon as it's released in the form of a mainstream commercial product. I can only hope that Ananova sounds as good as RealSpeak.

What other open source project would you expect Netscape Communications Corp (or AOL) to be involved with?

The fact that it has taken a whopping long time for the (marginally usable) M10 release to arrive is not a clear example of failure; the project has had to labour under several significant constraints:

• In order to release Mozilla as Open Source(tm), Netscape had to tear out a whole lot of code that they didn't own. Java, VisiBroker, RSA stuff, ObjectStore, TrueDoc, Full Circle Talkback, Inso Proofreader, and others.

  This left gaping holes in the source code tree, things that had to be reimplemented.

• Mozilla has essentially been rearchitected.

  What with the above gaping holes, and other things that had grown into being ill-designed, it made huge sense to rebuild a whole lot of the functionality from scratch.

If a version that is of "production quality" is released in the next 4 months, which is not inconceivable, that essentially means that Mozilla has been recreated in two years, which is certainly not a monumental failure.

I had an opportunity to speak with Dr. de Garis over a year ago at a party thrown by an acquaintance of mine who had interviewed de Garis for a documentary on Nanotechnology and AI. I found Dr. de Garis intelligent, personable and amusing.

At the time he was rather pessimistic about the Robokoneko project, but mostly because of the cultural problems he was dealing with as a Britisher in Japan. However he claimed that the artificial neuron work was proceding well, even though they were doing it all with simulators. He predicted then that, before 2000, they would be creating silicon versions. From the information in the links it would seem that his prediction has come true. Only they are using FPGA chips instead of going to a foundry for CAM specific VLSI.

It is interesting to note that Dr. de Garis has made incredible progress by following a path the mainstream AI community has largely discounted -- that of modeling real neurons and real brain structures. I wonder what will come out of his next collaborative development at Starlab in Brussels? From his statements to me I would certainly hope he would find the living and working arrangements more congenial.

I do find it very interesting that he will be working with Lernout and Hauspie (developers of Voice Recognition software). The spin-offs from that may be more important than the original research!

Jack