noise canceling?

[Wyatt]

I got an idea yesterday, so I decided to look into how noise canceling works.

I have been reading about it, that they use an inverted wave form.

Well, that sounds simple enough, but when I use an inverted wave form..

..it does almost nothing at all.    

Can you tell me what am I missing here?

THX

[kara]

Well, it's strange that it doesn't anything....
It's indeed how noise canceling works, as far as I know....
Taking a sample of the 'pure' noice, a trunc of the track where there is only noise, generating the inverted wave and mixing it in.

I've done the test in audacity...
See pic 1, I've created a sine wave and inverted it in the second track
Then in pict2 i mixed it.
Result : As you can see the result isn't zero, but I get a litle sine at -24 db

So, it should 'something'....

[Wyatt]

Well, it's strange that it doesn't anything....
It's indeed how noise canceling works, as far as I know....
Taking a sample of the 'pure' noice, a trunc of the track where there is only noise, generating the inverted wave and mixing it in.

I've done the test in audacity...
See pic 1, I've created a sine wave and inverted it in the second track
Then in pict2 i mixed it.
Result : As you can see the result isn't zero, but I get a litle sine at -24 db

So, it should 'something'....

OK..I will try mixing it..what I did here was to delete the right track of a stereo file..then copy the left track to the right track..then invert the waveform of the right track.

I could hear a difference, but not really quieter. Just a little distorted.

I will try mixing..

Thanks..I am hoping I can get this to work..

[kara]

Well if i understand right what you're saying, this wouldn't work because you send the inverted track to another channel. In that case it wouldn't cancel anything.

k

[Wyatt]

Well if i understand right what you're saying, this wouldn't work because you send the inverted track to another channel. In that case it wouldn't cancel anything.

k

That makes sense..I'll keep working on it..thanks!

[Wyatt]

Whoaoo!!!  ..does it ever work!!!!

I would like to be able to create a quieter environment in my truck. Imagine popping a CD in the player and having it get quiet all of a sudden.   

I am not terribly optimistic that I can succeed entirely, as I am thinking about this, because so many of the sounds fluctuate continuously..

..but I am very curious about it, so it is worth it to me to pursue it.

Thanks for helping me out this morning kara. I appreciate it.

You can see in the picture that the session is playing and the white cursor is on a fairly loud place, and at the very bottom left of the screen the level indicators are totally dead flat zero..not a twitch.

[Wyatt]

Not only does it not work in the truck, but it doesn't even work a little bit.

I got a preview of that while still in Audition. When I shifted the invert even the slightest bit, they just added together. 

..also I could not line up a different portion of the timeline successfully either.

This one has to be totally spot on. I see why they use a microprocessor to do it in real time.

I found a schematic for it, but anything the mic picks up would get canceled, including any conversation and also including any music you are listening to..so that's why it's headphones only.

Hmm..a speaker setup would make for a "quiet room", though..interesting. 

[kara]

When I shifted the invert even the slightest bit, they just added together. 

That sounds logical doesn't it?
Actually there are allways added but 1 an -1 ad up to 0
Imagine the following bitstream

1 0 1 0 1
Now the inverted : 0 1 0 1 0
if you do a logical AND on it it gives
0 0 0 0 0

Now shift it just one bit
you get
1 0 1 0 1
AND
0 0 1 0 1 0
and you get
0 0 1 0 1 0

NO canceling anymore 

k

[Oren]

The noise canceling function in Bose headphones is described as constantly sampling noise in the listening environment, and nearly simultaneously generating a reverse-image in the headphones, along with the music. Exterior environmental noise and interior generated noise cancel each other out, and only the music is audible to the listener.
 The process has to happen in real time, however, in order to be effective. A recording of "typical" in-Wyatt's-truck noise will not match the actual in-the-moment noise signature closely enough to provide noise cancelling.

[folderol]

Basically, what you are trying is 'whack-a-mole'!

For in-room suppression, except at very low freqencies the 'anti' sound source has to be as close as possible for both phase, and power distribution field reasons. Therefore if you have lots of noise sources you need lots of indipendent suppressors.

A company I used to work for (till they went bust) specialised in video conferencing kit, and they had some very sophisticated self-teaching echo and noise cancelling kit, but this relied on an extensive array of ceiling microphones and only performed the cancelling for the remote receivers (or recording system). Having said that was seriously impressive - so was the price!

[Wyatt]

When I shifted the invert even the slightest bit, they just added together. 

That sounds logical doesn't it?
Actually there are allways added but 1 an -1 ad up to 0

NO canceling anymore 

k

Yeah..thanks!  ..that makes perfect sense..I learned some stuff this morning and had a good time in the bargain!

The process has to happen in real time, however, in order to be effective. A recording of "typical" in-Wyatt's-truck noise will not match the actual in-the-moment noise signature closely enough to provide noise cancelling.

Yep..the in-truck noise sounds kind of "average". but it's not..it's very specific..when I put the waveforms under the microscope it became clear. Even so, I tried it out in the truck, since Harley wanted to go for a ride anyway.   

I was going to town the other day, and some high pitched wind sounds did a remarkable job of masking some of the higher freqs of one of my songs that gets gradually higher pitched as it goes along. That's when I started wondering about Canceling..I vaguely remembered hearing about the Bose 'phones, but didn't know what was going on.

That's what got me started reading and playing around with it this morning.

Basically, what you are trying is 'whack-a-mole'!

Bingo!!  ..LOLOLOL..now that really describes it. 

For in-room suppression, except at very low frequencies the 'anti' sound source has to be as close as possible for both phase, and power distribution field reasons. Therefore if you have lots of noise sources you need lots of indipendent suppressors.

A company I used to work for (till they went bust) specialized in video conferencing kit, and they had some very sophisticated self-teaching echo and noise cancelling kit, but this relied on an extensive array of ceiling microphones and only performed the cancelling for the remote receivers (or recording system). Having said that was seriously impressive - so was the price!

Awesome stuff. I felt like I was in a laboratory this morning, and I appreciate everybody's help. This could have been pretty frustrating, especially when my first logical attempt was such a failure.   

[Wyatt]

Final note:

One millisecond difference between the two tracks is enough to nullify the canceling effect.

[kara]

Final note:

One millisecond difference between the two tracks is enough to nullify the canceling effect.

Based on the maths, it should be less then one milisecond... depends on the samplerate of course

k

[Wyatt]

Final note:

One millisecond difference between the two tracks is enough to nullify the canceling effect.

Based on the maths, it should be less then one milisecond... depends on the samplerate of course

k

I'm sure you're right.

I woke up thinking I would try 1 ms..I didn't go any smaller.