Page 1 of 2

Command Voice Interface

Posted: Sun Feb 26, 2017 11:59 am
by Interrogative
Instead of SIRI or Cortana, why can't we have a spoken AmigaDOS interface, or CVI? It would function much the same as the CLI except that you would be speaking the commands, options, and paths into a microphone. With enough scripting (ARexx, Python, or AmigaDOS) perhaps it could be a little like SIRI?
I am certain that there are Amiga users who have been whining for AmigaOS to drop keyboards and mice in favor of a touchscreen interface. If AmigaOS goes down this route, I ask that it keep legacy support for physical keyboards and selection devices. AmigaOS may tap into a market of people who suffer from zombie finger, who hate touchscreen interfaces, and who might just like a well designed user interface. If Hyperion does ditch the traditional Amiga user interface, I ask that they skip over touchscreen interfaces for a CVI (Command Voice Interface). Instead of a CLI (Command Line Interface), let the user speak the commands and paths to the Amiga, which will respond via a new version of the narrator device.
https://royceonarampage.wordpress.com/2 ... gy-trends/

Re: Command Voice Interface

Posted: Mon Feb 27, 2017 1:55 pm
by Belxjander
Additional Input mechanisms as substitites for the keyboard/mouse/game devices we can currently use?

So instead of Siri/Cortana/... We use a "Sabrina" or other trigger name?

Re: Command Voice Interface

Posted: Tue Feb 28, 2017 12:02 am
by tonyw
I think someone has been working on a voice-recognition program, but as a private development. I can't tell you any more than that.

But you can imagine that the AI is the major part of what you are asking for. The interface to the user (perhaps I should say the owner, since the AI may become the user) is trivial in comparison.
When you think of the resources that Google, Microsoft et al are pouring into these projects, it's not worth our while even thinking about it.

Re: Command Voice Interface

Posted: Tue Feb 28, 2017 8:46 pm
by trixie
I think this forum badly needs a pipe-dream filter :-)

Re: Command Voice Interface

Posted: Wed Mar 01, 2017 4:47 am
by LyleHaze
trixie wrote:I think this forum badly needs a pipe-dream filter :-)
In some ways, I agree completely. We have limited resources, and it's important to focus on things that will give us the best return.

But on the other hand, this is a VERY special OS, and we need to remember that dreams can sometimes lead us into realities.


Google and Amazon are battling for the turf, I think we are out distanced already.

But I will always CONSIDER anything posted, just in case there's something amazing hidden there.

:)

Re: Command Voice Interface

Posted: Thu Mar 02, 2017 2:58 pm
by Daedalus
While a complete voice recognition suite like Siri etc. is a mammoth job, something closer to "voice tags" is far more feasible, where the software doesn't try to translate speech into text, but instead matches sounds it hears to a list of known sounds and reacts accordingly.

VoiceAttack is a present-day bit of software for PCs that uses this model, and even the Amiga had its own VoiceShell back in the day. I used it on a stock A1200 with a parallel port audio sampler at the time, and while it sucked up about half the A1200's CPU time, it worked reasonably well - well enough to use it for controlling a relay card via ARexx.

Saying "Computer: Lights on" to turn on the lights in my room was hella awesome when I was a teenager in 1998 I can tell you :D

Such a utility could easily be implemented on any OS4 machine if someone was up for it. 68k ASM source is included, but that's of limited use really.

Re: Command Voice Interface

Posted: Fri Mar 03, 2017 12:18 pm
by Interrogative
Daedalus wrote:While a complete voice recognition suite like Siri etc. is a mammoth job, something closer to "voice tags" is far more feasible, where the software doesn't try to translate speech into text, but instead matches sounds it hears to a list of known sounds and reacts accordingly.

VoiceAttack is a present-day bit of software for PCs that uses this model, and even the Amiga had its own VoiceShell back in the day. I used it on a stock A1200 with a parallel port audio sampler at the time, and while it sucked up about half the A1200's CPU time, it worked reasonably well - well enough to use it for controlling a relay card via ARexx.

Saying "Computer: Lights on" to turn on the lights in my room was hella awesome when I was a teenager in 1998 I can tell you :D

Such a utility could easily be implemented on any OS4 machine if someone was up for it. 68k ASM source is included, but that's of limited use really.
The CVI I had in mind would recognize spoken AmigaDOS commands and paths. I seem to remember an Amigaworld (magazine) article in which a blind man used a Video Toaster along with some voice recognition and the Speak device to make videos. That is the kind of thing I had in mind. If it could be done on the Amiga in 1993, why not now?

Re: Command Voice Interface

Posted: Sat Mar 04, 2017 8:42 am
by LyleHaze
I once set up an A500 for a friends blind sister. She had found a typesetting program that was intended for use by the blind.
To make basic computer operations easier, I set up a shell that had all STDOUT and STDERR re-directed to the translator/Narrator.

It worked amazingly well, with the biggest "Oops" being the super long floppy formatting times required because it was narrating the progress of every track.

She worked that system as a typist for local Doctors, typing from audio recordings they provided. Whenver I went over to help out, I'd have to remember to bring a monitor.

Good times, long ago.

Re: Command Voice Interface

Posted: Wed Mar 08, 2017 4:52 pm
by Flipflop
There is an old one resp. similir thing available in AmiNet called VoiceShell or the even older stuff from Richard Horne, but cant tell if its working within OS4.x.

Re: Command Voice Interface

Posted: Fri Jun 23, 2017 11:22 pm
by Belxjander
Well we can just use a minimal "trigger recognition" tool and pipe the rest of the command processing for "speech to text" through the Google/Microsoft API's online when it is available...

Let them do all the heavy lifting and deal with the *text* results (possibly a different online service?)

*might* be practical for those systems that have reliable network connections at least?