noise cancellation circuit

Thread Starter

cycleops

Joined Dec 4, 2011
3
Hi,

I'm looking for advice on the suitability of using this noise cancellation circuit to (hopefully) improve speech recognition app running on my home theater pc:
http://gilmore2.chem.northwestern.edu/projects/noise_prj.htm

I thought of using the audio out connections of my receiver/amp as input to J1 and the mic as input to J2. The mic would be about 12 feet from the speakers.

Does anyone think this would, or would not, work?

BTW, I am an electronics beginner.
 

SgtWookie

Joined Jul 17, 2007
22,230
What is it that you are trying to do?

Perform speech recognition while playing a movie?

It might help cancel a fair bit of the noise, but will likely be less effective in the higher frequencies due to the propagation delay (lag) between the line-level output, and output that goes through the amplifier, speakers, and then acoustic sound that travels through the air 12 feet to the microphone.

That circuit has only a very small amount of propagation delay from the mic input to the cancellation of noise.
 

kubeek

Joined Sep 20, 2005
5,794
Hi,

I'm looking for advice on the suitability of using this noise cancellation circuit to (hopefully) improve speech recognition app running on my home theater pc:
http://gilmore2.chem.northwestern.edu/projects/noise_prj.htm

I thought of using the audio out connections of my receiver/amp as input to J1 and the mic as input to J2. The mic would be about 12 feet from the speakers.

Does anyone think this would, or would not, work?

BTW, I am an electronics beginner.
I don´t see how are noise-cancelling headphones gonna help with recording sounds for speech recognition. Subtracting speaker signal from the mic won´t help at all.

With headphones it helps because the path between the speaker and mic is very short.
To do the same for recording, you would have to shove a mic into your mouth, which defies recording with another mic.
 

praondevou

Joined Jul 9, 2011
2,942
I hate when they say "noise cancellation" when it's actually noise attenuation. If there only was such a thing like complete noise cancellation. Not even they most expensive headphones manage to cancel noise...
 

Thread Starter

cycleops

Joined Dec 4, 2011
3
What is it that you are trying to do?

Perform speech recognition while playing a movie?

It might help cancel a fair bit of the noise, but will likely be less effective in the higher frequencies due to the propagation delay (lag) between the line-level output, and output that goes through the amplifier, speakers, and then acoustic sound that travels through the air 12 feet to the microphone.

That circuit has only a very small amount of propagation delay from the mic input to the cancellation of noise.
Yes, I do want to perform speech while watching movies (pause, volume, play, etc...). I'm using voxCommando (http://voxcommander.com) and it works surprisingly well, but I'm hoping to improve its success rate.

To clarify for kubeek, there are no headphones involved with my intended use. The circuit would invert the signal from my receiver/amp and apply it to the signal from a mic sitting on a shelf. So the circuit attenuates sounds from the amp (like a CD I'm playing), and more clearly hears my speech via the mic (like me saying "pause music"). It feeds my speech into my home theater pc where voxCommander recognizes and acts on the speech.

I'm wondering if the lag SgtWookie mentions will prevent attenuation.
 

kubeek

Joined Sep 20, 2005
5,794
The mic and the speakers would have to be bolted to floor for this to work at least a bit with a simple circuit.
Unfortunately the direct wave from the speaker mixes with the reflections off all surfaces. You would need two circuits which delay the L and R speaker signal and mix it with the mic signal, but I doubt if will have any cancelling effect.

I can suggest you one thing, take a piece of music or anything, play it through one speaker only and record the result, without any other noise. Then put the original singal and the record into software editor, for example cooledit would work nicely, and try your best at combining the two (addition or subtraction + delay, maybe some equalizer) and see if you can make anything out of it that has at least lower volume than the original mic signal is, i.e. try cancelling the two to get zero signal.

If you succeed, then you will have the recipe how should the circuit look like to get the result. Much easier than blindly building and testing. Also then try recording again and see if it still works :)

P.S. you can add multiple copies of the original to tame the larger reflections
 
Last edited:

Thread Starter

cycleops

Joined Dec 4, 2011
3
The mic and the speakers would have to be bolted to floor for this to work at least a bit with a simple circuit.
Unfortunately the direct wave from the speaker mixes with the reflections off all surfaces. You would need two circuits which delay the L and R speaker signal and mix it with the mic signal, but I doubt if will have any cancelling effect.

I can suggest you one thing, take a piece of music or anything, play it through one speaker only and record the result, without any other noise. Then put the original singal and the record into software editor, for example cooledit would work nicely, and try your best at combining the two (addition or subtraction + delay, maybe some equalizer) and see if you can make anything out of it that has at least lower volume than the original mic signal is, i.e. try cancelling the two to get zero signal.

If you succeed, then you will have the recipe how should the circuit look like to get the result. Much easier than blindly building and testing. Also then try recording again and see if it still works :)

P.S. you can add multiple copies of the original to tame the larger reflections
I think you're right, there would be no canceling. Thanks.
 
Top