Open Source Speech Recognition - With Source

Paul Lamere writes " This story on ZD-Net and this recent story on Slashdot describes the recent open sourcing of IBM's voice recognition software. This release, unfortunately, doesn't include any source for the actual speech recognition engine. Olaf Schmidt, a developer on the KDE Accessibility Project , is quoted as saying 'There is no speech-recognition system available for Linux, which is a big gap.' In an attempt to close this gap, we have just released Sphinx-4, a state-of-the-art, speaker-independent, continuous speech recognition system written entirely in the Java programming language. It was created by researchers and engineers from Sun, CMU, MERL, HP, MIT and UCSC. Despite (or because of) being written in the Java programming language, Sphinx-4 performs as well as similar systems written in C. Here are the release notes and some performance data."

Comments

Aim You Sing Ate Write How

[Anonymous Coward]

Ate lurks barry wall.

Re:Aim You Sing Ate Write How

[BarryJacobsen]

Ate lurks barry wall.

Who ate my wall?

Virtual Machine Syndrome

[nihilogos]

Colloquially known as "pointer-envy", this condition may affect all programmers, but is especially prevalent in java and C# developers. It is most easily recognized in a release announcement, where for no reason whatsoever the afflicted developer suddenly interjects a statement like "and it's just as fast as C", to the bewilderment of the audience.

Treat suspected cases with caution, and under no condition contradict the patient. There is no known cure.

Re:Virtual Machine Syndrome

[Xeger]

KNOWN CAUSES: Recent research results from information-theoretic psychoanalysts shows that Virtual Machine Syndrome is most likely a pre-emptive defensive discourse strategy. VMS sufferers typically become symptomatic after months or years of constant haranguing at the hands of colleagues, friends and professional contacts that anything they write, regardless of its execution environment or portability requirements, could have been done "better and faster in C." Oftentimes, such criticism is levied against VMS sufferers even when the application in question is I/O-bound and spends 80% or more of its time suspended, waiting for network or disk I/O to complete.

TREATMENT: Implement reliable and efficient systems using virtual machine of choice, regardless of criticisms. Apply free-market therapy judiciously, allowing adopters of Virtual Machine technology to thrive and become prosperous if warranted. VMS symptoms typically disappear when sufferer's stock options are valued at 300% of their strike price. Symptoms may also be temporarily relieved through just-in-time compilation.

RELATED SYNDROMES: Ossified Self-Important Myopia (OSIM), which is the tendency to assume that one's favorite programming paradigm, language, or OS is unconditionally and unreservedly the best choice for any software project. Characterised by the inability to understand that the only way to guarantee maximum efficiency is to write everything in assembly language, with complete and perfect knowledge of all quirks of the specific target instruction set.

Re:Virtual Machine Syndrome

[Brandybuck]

Try my new text editor, it's written in Java!

Why should I?

Because it's written in Java!

How is it better than what I'm currently using?

It's written in Java!

I'm already using vi, emacs, kate and gedit, why should I use yours as well?

Because it's written in Java!

Does it have a spell checker, syntax highlighting, and auto-indent?

Who cares? It's written in Java!

Name two benefits to your text editor?

That's easy! First, it's written in Java. Second, it's uh... uh... hang on, uh... it's written in Java! Yeah, that's it, it's written in Java!

Re:Virtual Machine Syndrome

[IgnoramusMaximus]

Check out "Metronome"

I dont give a damn about Metronomes and speech recognition of questionable usefulnes. None of those are Java apps I deal with. And of those I deal with all suck.

worm-friendly apps?

Speaking of security, most Java apps are deployed in places that need them not in the first place, as a kludge for an E-Commerce site or electronic banking interface which can be done with a bit of thinking in plain HTML. Others, like IBMs for example, are mainly administrative tools which have no communication abilities outside of their narrow scope. These, if made in any other language would not be any more prone to worms. As a matter of fact, the use of Java on some of these electronic commerce sites introduces unneeded complexity and results in code executing on customer's computers whereby they become prone to being abused by spoofed/buggy VM's etc.

So, great story, but not exactly correlated with reality.

Reality? Oh dear. Listen dude, I am telling you as a user of your wonderful computer science masturbation effort otherwise known as Java: No. Nada. Niet. It aint a go. No can do. The bank we deal with is rewriting their apps to be java-free because of the amount of flaq they are getting (and no they are not going to that other aberration known as C# either). IBM DB2 is banned in many companies we deal with. Etc etc. We, the users, not you, Mr. Java Wanker, have the final word on this. Trust me.

Re:But what about text to speech?

[Sheetrock]

Given that there is already a rudimentary text-to-speech package available for Linux, and now a speech-to-text package, perhaps the secret is to pipe one to the other in a closed loop until one learns how to enunciate and the other how to listen?

Open Source - With Source!

[NSash]

From dept-of-redundancy-department?

I'm not one to be picky about titles, but sheesh...

Obligatory Commercial Break

[Noksagt]

Woman: [dictating into cell phone] To: Mike. I had fun last night.
Cell Phone: To: Mike. I have lip fungus.
Woman: [into cell phone, angrily] I had FUN, not lip fungus!
Cell Phone: I have fungus, not lip fungus.
Woman: I DON'T HAVE LIP FUNGUS!!!

Obviously

[Moderation abuser]

"This data was collected on a dual CPU UltraSPARC(R)-III running at 1015 MHz with 2G of memory."

Looking at the performance data it just blazes along on that config. Not exactly what I'd call an embedable system, though Microsoft might beg to differ.

Re:The Myth Must Die

[nihilogos]

many benchmarks have shown that a modern optimized JVM with JIT compilation is roughly equivalent with most implementations of C++, with some benchmarks being better for Java and some being better for C++.

And many studies have shown that going with Microsoft software is cheaper than going with open sourced software.

Re:Java!?!

[darkonc]

Quick someone port this to C.

Just be glad it wasn't written in Lisp.