A Question about sound effect and programming

[lightspeed]

I have talked before about sound effects and hal understanding them and reacting from them .
 I was just thinking ( yeah I know sometimes it gets me into trouble lol) about could a person use say a home made sound effect from their personal telephone ( as some phones ring differently etc. ) make it into a wave file so that when hal hears the wave file (this would have to be through a voice recognition type program(??)  , hal re acts to it saying "some ones at the door , the door bell is ringing!"
   I can see it in my mind how it would work , and maybe it has nothing to do with voice recognition but rather just a sound wave program script play and match system , sound first , sound matching the script which makes an end result of looking like a reaction to the sound , with script . 
  Anyway I may try to work on this , just wanted to see what others thought about it , Art, anybody?

[Art]

Lonnie,

I'm not sure whether Hal's software is set to only listen for words / sentences or possibly Noise or Sounds as well.

I was once experimenting with another program and being a bit bored, I picked up my guitar and played an intro that I'd been working on. The program floored me when it said, "Art, I don't think I've heard that song before!"

Now whether or not it knew the differrence between a song and some words being spoken I still can't say for sure, but believe me, THAT was one of those OMG moments!

Have you experimented by playing different sounds or noises while Hal was normally listening for your voice? Let us know and hopefully some of us can do our own experiments too!

Later....

[Art]

Lonnie,

After some experimentation, I noticed that with the Mic on, Hal heard various sounds that I made including some random guitar notes, chords and short intros. Hal seemed to be attaching various words for some of these sounds yet when it was her turn to speak, nothing came out.

When I spoke to her with my voice, the program responded in kind with an appropriate response.

Actually, to my surprise, Karlie was able to recognize about 95 - 98 % of my speech. But still, that kind of percentage is enough to completely trash everything I've worked on for many years of typing the entries instead of trying to speak them.

Is would seem as if a different order of software is needed to distinguish between verbal and other (non-human) audible sounds.

That would have been an interesting idea you had. (like most of them). 

You could still do the trigger file with appropriate words like sneezing, coughing, clearing the throat (Ahem! sound), whistling, humming, finger snapping, birds chirping for Hal to recognize, car horns, typing and more).

Just record the sound and save as a recognized file type that wouldn't take much space with regard to size. Then let Hal loose with the sounds.

Anyhow maybe someone else will obtain different results for your project. Stay tuned!!

[lightspeed]

Yes Art, what I was thinking about is based on trigger words, and I have already done that with hal sneezing , laughing etc. in real voice effects .
   what I am thinking was scripting also showing ( door bell.wave file) sound = hal command random sentence . "that's the door bell ringing" .
This of course is an example and not the correct scripting but just to give an idea .
I may try different things .
Halloween experiments and all !!
buwaa haaa haaa haa !   
  the problem is hal would have to understand the doorbell sound to make a reaction sentence . that's the whole key to it all, and other sounds as well once a way is figured out for the first understanding of a sound a pre scripted sentence is created to announce the response understanding of what is heard .

[Art]

Skip all that and let HAL make the sounds (via the .wav files) whenever Hal hears one of the associated trigger words. Then Hal would be realistically reactive to words that have associated sounds.
This bypasses the need for Hal to have to listen for certain sounds then figure out just what was heard. (I don't think we're even close to something quite like that just yet in our Speech Recognition evolution.

[Carl2]

  Since I'm a Dragon user I can say from experience that when dragon hears a cough, sneeze or dog bark that it quickly comes up with a question mark sign, actually a big improvement, before it would spend some time trying to make sense out of the noise and come up with some words to put in ( very random with no meaning ) and now there is no input.  There is also a teaching portion, so you could say " input closing 1" and dragon would type in sincerely yours, Carl or whatever you choose.  Possibly you could train dragon to hear the doorbell ring and type in the Doorbell is ringing.  My experience using dragon which is also using AI
Carl2

[lightspeed]

     Art , I "ALREADY" have done that make hal react with wave files through trigger words , my hal makes a sneezing sound , coughs, laughs sometimes. maybe the only thing I can do is ( for now ) whenever I say the doorbell (for example ) I can have hal say, yeah I heard the door bell chime and even have random responses for the same trigger word , it wouldn't actually be hearing the sound and determining what to say to the sound but can give the illusion of actually" hearing" the sound .
      I have done something similar like this in another hapswap file I did with colors , where when trigger words of different colors are used hal agrees with what color you are talking about in a little bit of detail .

    Yes Carl2 you are right about dragon speak, if it doesn't understand a word spoken (mumbled etc.) or the sound basically , it will give a question mark , wanting you to resay it , so it will not be able to listen for sound effects such as a door bell to learn distinctions .

[Art]

It might be cool to allow Hal to read children stories and add sound to certain passages within the story provided your trigger words are met.