Speech Recognition?

Blitz3D Forums/Blitz3D Programming/Speech Recognition?

John Blackledge(Posted 2005) [#1]
I found an old thread about speech recognition for Blitz3D, downloaded all the files, and it totally failed to initialise for me as it did for others on the thread.

Does anyone know if this was ever finished, or if there is another (Blitz3D) method?


Andy(Posted 2005) [#2]
Link please?!

Andy


John Blackledge(Posted 2005) [#3]
http://www.blitzbasic.co.nz/codearcs/codearcs.php?code=951
Cheers,


jfk EO-11110(Posted 2005) [#4]
When you're using Win XP, make sure "MS Agent" is working. Then the SAPI should be installed. SAPI Speech IO is part of the Agent.


John Blackledge(Posted 2005) [#5]
Well, I've installed all the required downloads from MS, but there is no where in XP that I can find any sort of test of the Agent.

Control Panel > Speech Properties shows that it is installed, and all the helpfiles describe what the system is, but nowhere can I find how to access it.

There is no example!


jfk EO-11110(Posted 2005) [#6]
Isn't there a folder with some scripting host example scripts? (.js and .vbs)? Did you try to search for the string "speech" in filenames on the harddrive?

Additionally, there's a massive Agent "Modding" community where you can download aditional characters.


John Blackledge(Posted 2005) [#7]
To the first question, no, they are just a bunch of exes that you run and install. No examples. (Not like MS at all.)

To the second question, I don't want an agent at all.

Beeps obviously wrote some code which he got working, but from the look of the thread above almost no-one else did.

What I'd like is for Beeps (Binary People appear to not exist any more), or anybody, to explain how to use his code.


jfk EO-11110(Posted 2005) [#8]
Sorry. I just remember some years ago I downloaded the entire SAPI (dunno, 5.something or so) SDK, a 100 Megs Download, and installed it in Win98se. I was capable of running .js script from Darkbasic. There was also a Speech recognition example with training etc.

I never used the one of beeps. I just mentioned the Agent because if you are capable of using an agent, you'll know speech is properly installed and configured.


Damien Sturdy(Posted 2005) [#9]
not quite JfK. on XP, agent is installed but speech wont work until you get the files frmo MS website

You'll have to poke aronud quite alot, but search www.Microsoft.com for "Microsoft Agent". It'll give you a ton of information. I dont have any links im afraid.

In windows 2000, Agent and the speech API worked fine, and the characters would talk. I found that in XP, i could get them to appear but not listen.

Best of luck :)


John Blackledge(Posted 2005) [#10]
Thanks, guys. Actually I've done all that.

I've downloaded the agents, and installed them.

And nowhere in Widows XP can I find out how to see them, trigger them, or use them in any way.

To all intents and purposes they still don't exist, even though the speech engine is shown as installed in Control Panel.

If I didn't know better I'd think this was all a scam by Microsoft.


Damien Sturdy(Posted 2005) [#11]
I wish i had smoewhere to host, at hmoe, i got a prog you can test i made with VB, its like a desktop pet. If it installs and works, you're dandy and its all there.
if not? Poop.!
Did the MS Agent Test webpages work?

[edit]
I should probably point out that windows XP does have speech installed, just not the one that works with Agent for some reason :/

I wonder....


John Blackledge(Posted 2005) [#12]
"Did the MS Agent Test webpages work?"

Ah, yes, I just remebered the 'test' button.
Nothing happened. ?


BlackJumper(Posted 2005) [#13]
AFAIR: (disclaimer !)

The dll posted by Beeps was effectively a 'build' of some SAPI 5 functions wrapped using some other programming environment (e.g. VC++) The original SAPI5 SDK (download from Microsoft) had a load of examples. Beeps provided a couple of samples that (1) showed how to use TextToSpeech and (2) showed how to wait for voice activation and then return the input as a string.

I did a bunch of this stuff myself using BC++ a while back, but couldn't get it to recognise French phrases unless delivered in an american accent, so I gave up on it. When I returned to it a while later MS had gone heavily over to Server-based Speech applications and the whole MSAgent-based approach. It seemed to be a whole new ball game, so earlier libraries and code simply didn't work anymore.

I have just checked, and I have a folder I have named "Superceded Beeps SAPI" in my decls folder, so maybe an earlier version of the dll might still work. email me if you want a copy of that folder.


John Blackledge(Posted 2005) [#14]
Thanks, BlackJumper, for all the above info, the email is on it's way.

I must say that I'm left with a feeling of frustration now, that the technology is all out there (how many thousands of man hours must MS have put into it?), but it just does not seem possible to get my hands on it.

From what you're saying it would seem that I have to download a (and learn) a whole new SDK!

If you contrast that with the user-friendliness of the MCISendString library (also MS, see my other thread) which also did a pretty complicated job but by a very simple method, then you definitely get the impression that different teams produce different section of Windows.

This may sound like a truly obvious statement, but I'm talking more about intention and implementation.....

There was a move during the 80s to rid the industry of the 'men-in-white-coats' image.
Now MS seem hell-bent on reinstating that image; in other words facilities like this are 'only for professional developers' - mustn't have the man in the street actually being able to use it.
I've been with Windows almost from the beginning, and you do begin to pick up a change of attitude towards developers at large on their part.

When you contrast this with what Mark Sibly has achieved (yes, I felt that this thread was getting too far away from Blitz too!) in putting an unbelievable and amazing 'wrapper' around DirectX.... well, need I say more.


BlackJumper(Posted 2005) [#15]
Earlier version sent... hope it helps.

For the record - the new MS way of doing things seemed to be an attempt to give the end-user a simple way of compiling dictionaries (via XML) and customising speech agents (using wizards ??) There was an element of 'white-coatishness' in the server side of things - probabaly a necessity in the web-enabled, distributed-processing, telephony-leaning environment that the company were gunning for.

... Pity it pissed all over about 8 months of a project I had been working on !!


John Blackledge(Posted 2005) [#16]
Yeah, let's take the opportunity (and then drop it, this is the Blitz forum after all!) to rant and rave about MS moving the goalposts yet again.

Why am I not surprised. My sympathy to you.

Hang on, I'm pretty sure that the original selling point by which Bill Gates got the deal with IBM was the promise of total backwards compatibility each time the OS was changed???!!!

Hey, Bill, what happened?
Or didja just lie?

Even _I'm_ getting fed up with this thread now!
Life's too short. Lets' move on.