controlling with voice

[Abagtha]

Hey friends...

I am an Electrical and Electronic Engg. student. My friend and I have to make our final year project.

I would like to know if anyone can help me with a circuit on controlling with voice.....say...the tubelight or radio or any small gadget with voice. May be something like...when i say "on" the tubelight gets switched "on" and when i say "off" the tubelight gets switched "off".

This is just some idea i have in mind. Do you have any other better or feasible ideas..?

Thanks....
Abagtha

 

[audioguru]

Are you allowed to cheat a little?

Say, "Turn on." and it turns on.
Say, "Off. (or Stop.)" and it turns off or stops.

Hint: It counts the bursts of sound for each word. Two bursts turns it on, one burst turns it off or stops it.
Don't let your teacher see a dog controlling it. :lol:

 

[Someone Electro]

Yea.

Even real speach reconition is NOT reliabe even on a PC.I once tryed one of those apps that you can control your PC whith.It constantly mistakes noises for comands and reconises a word wrong.You have to speak realy slow and reay realy realy clear.

Not a good project to go on.

 

[Roboticinfo]

Well, if you only need to things Bernie it is quite accurate and company's use one chip that is at the top of the line right now.

Go to this website and see which one you want

https://www.sensoryinc.com/

Bernie you most likely have not tried the new dragon naturally speaking or you have never used a kit from sensory inc.

They mostly never mess up now it is like a 95% to 98% accuracy now.

I have used the kit maybe 4 times and it has never let me down yet well maybe twice or so but that is it.

 

[mstechca]

the two things you can detect from voice (if done electronically), is the duration of a syllable and the volume. Other than that, sophisticated circuitry is required. I don't do speech processing circuits, so I can't help you any further.

 

[roboticmisinfo]

I imagine if you are doing this for a Uni project, then it will be getting the principles right, rather than making something that is 100% reliable. I have never built voice recognition circuits myself, but I imagine it would be very difficult designing your own. I reckon Audiogurus suggestion was pretty good. If you could design something similar to a VU meter on a Hi-fi, then have it triggering at a set Voltage. You could use a LM3915 and have one of the higher level outputs as the trigger. I'm not sure if this is the right advice, as I am only a beginner (someone please tell me if I am wrong!!!)

Take a look here:

**broken link removed**

By the way. If you are just turning a tube light on and off, why not just have it triggering on one syllable for both on and off. You can't turn a light on, and then turn it on again without turning it off can you? Unless you are doing anything more complexed, I don't see why you can't do something similar to one of those "clap" lights where you clap your hands to turn it on, and then clap them again to turn it off.


Sorry if my ideas are stupid. I'm still learning,

Kind regards,

Mike

 

[audioguru]

Hi Mike,
The LM3915 contains 10 comparators and a bunch of other un-neccessary stuff, for a fairly high price. To detect a voice a circuit needs only a single cheap comparator, or just a rectifier charging a capacitor. :lol:

 

[roboticmisinfo]

OK, sorry about that, I realise now it was a pretty daft idea. So you mean just an op-amp comparing the signal to the signal you want it to switch at? e.g. having a set voltage of say 0.8V, comparing it to say a 0-1V signal from the mic, and then having the op-amp output going high when the input voltage goes above 0.8V? Sorry,

Kind regards,

Mike

 

[audioguru]

So you mean just an op-amp comparing the signal to the signal you want it to switch at? e.g. having a set voltage of say 0.8V, comparing it to say a 0-1V signal from the mic, and then having the op-amp output going high when the input voltage goes above 0.8V?

Yes, but a comparator switches its output much quicker than an opamp, if it matters. :lol:

 

[roboticmisinfo]

OK - Could you then just have the output of the op-amp / comparator switching a transistor/relay on & off for your lighting circuit? Like I say, I am sorry, I've just started learning

Mike

 

[Abagtha]

Thanks

Hey all...

Thank you very much for all the ideas and suggestions.

I am still deciding upon the project. My friend (with whom Im supposed to do the project), is quite interested in doing something with radio wave communications. So...we are tryin to see if we can do anything in that line as well.

Pl do help me out with ideas and suggestions. Thank you all

Regards
Abagtha

 

[audioguru]

OK - Could you then just have the output of the op-amp / comparator switching a transistor/relay on & off for your lighting circuit?

Actually, the output of a mic preamp is AC, so an opamp/comparator would switch for every half-cycle of the audio. The audio would need to be rectified and filtered to drive the input of an opamp/comparator selected for low DC input voltages.
Then the relay would be activated only while the sound is louder than the circuit's threshold, so the opamp/comparator would need to drive a flip-flop circuit.

It is a big difference from voice recognition:
One bark and the thingy turns on.
Two barks or another bark turns it off. Hee, hee. :lol: :lol:

 

[Someone Electro]

Just kill all the dogs in the area.(A shot gun shod work)

 

[roboticmisinfo]

Good idea. I never liked dogs anyway!! :lol: :twisted:

 

[roboticmisinfo]

So how would you make your own "voice recognition" circuit (rather than "sound activated") I thought you would have to have a lot of components/PIC's etc.

 

[audioguru]

My cable TV/ISP has a new voice-recognition attendant on the phone. After anything I say, there is a long delay before it switches me to the wrong place! :lol:

 

[Emantra]

Well I'm working on my Uni project about the voice activated door opener. On which I've written code for the speech recog. in MATLAB
there is some problem left in MATLAB which I'm unable to solve.

So the problem is that my code doesn't read the database where I've maintained two database to stored sound of CLOSE and OPEN from different ppl.

And the other problem is that I'm not able to link my parallel port interfacing code to my main program.

Can u guys help me on it?

 

[Someone Electro]

Microchip sells voice recognition libarys for there dsPICs.But by looking at the pdf i found it prety complicated.

And it probobly dosent work wery well since the dsPIC runs at "only" 120Mhz.

Its esyer to make voice.Since there are a lot of ICs out there that do it prety simply becose evryting is in the IC.You just send it a word whith an MCU and it says it out.