I think re-writing this as a nice commodity would be much easier. The algorithm used in voice.library is a simple pattern matching, not sophisticated speech recogntion.

_________________
--
Author of
HD-Rec, Sweeper, Samplemanager, ArTKanoid, Monkeyscript, Toadies, AsteroidsTR, TuiTED, PosTED, TKPlayer, AudioConverter, ScreenCam, PerlinFX, MapEdit, AB3 Includes and many more...
Homepage: http://www.hd-rec.de

@Hypex

Thanks a lot for offering a hand, but what I need most is your right brain

I've archived and uploaded all the files needed here. If you extract it somewhere and execute the AssembleVoiceLib script from the VoiceLibrary drawer it should compile just fine.

_________________
AlexC's free OS4 software collection

AmigaOne XE/X1000/X5000/UAE-PPC OS4 laptop/X-10 Home Automation

@Wanderer

VoiceShell already is a commodity

It basically provides the front end between the user and the library.

I think the original intent for the library was to make it easy to add voice command support directly in any program rather than having to load a specific dictionary in VCLI/VShell and then call ARexx commands to control the program.

I have nothing against merging voice.library and VoiceShell into a single application, as it would still do what most people need anyway.

The VoiceShell asm source is available on Aminet (VoiceShell.lha) if you want to look into it, but I might have to get you some includes from the author as it needs some file named "2.0.xref" which I couldn't find anywhere online.

Or do you mean creating a new commodity from scratch?

_________________
AlexC's free OS4 software collection

AmigaOne XE/X1000/X5000/UAE-PPC OS4 laptop/X-10 Home Automation

@AlexC

Hehehe.

Okay good job yes it assembles here. I will take a look at the files.

I was going to suggest commenting out the functions using AHI and XDEF them in the code. (External definition IIRC.) Then you could replace them with a C version you compile into an object file. And also "extern" the functions so the linker sees it. But it may be too much work and might just be easier modiying the assembler source.

@Hypex

Great, let me know how it goes and if there's anything I can do.

_________________
AlexC's free OS4 software collection

AmigaOne XE/X1000/X5000/UAE-PPC OS4 laptop/X-10 Home Automation

@Hypex

*bump*

/me waves a dusty $100 bill at Hypex

_________________
AlexC's free OS4 software collection

AmigaOne XE/X1000/X5000/UAE-PPC OS4 laptop/X-10 Home Automation

@AlexC

Quote:

waves a dusty $100 bill at Hypex

This has happened before!

Sorry, I have been thinking about it, in the London Beat way.

I downloaded the code and took a brief look. But I havent taken yet taken a deeper look. To see how much needs modifcation.

I got too invovled with other projects that are hlding me back. Perhaps time for a break from those.

Um, seeing at narrator.library uses 3 formant, 9/14 value phonemes, how difficult would it be to update voice.library to use 4 formant enevelope detection with sub/harmonic timbre approximation and intensity/frequency correction functions?

It seems strange to try to analyse a portion of speech at teh sample level, when the samples are only low quality due to humans only having 7 bit output at 8khz at the best of times?

Apparently MIT found out you can reorder teh samples in up to 100ms size chunks, and as long as the frequencies are there, the phases can be whatever you want, and the brain can hear it the same. What happens to computer recognition currently when you do the same?

_________________
The older and more respected a scientist is, the longer it takes to prove him wrong.