Building A noise cancelling program.

[erickulcyk]

Hello,

If there is a fixed noise like a faucet running, and I sit in a deskchair at a fixed location in an adjacent room, do you think it would be possible the build a program for either the desktop speaker (also fixed), or another computer somewhere else in the room which would cancel out the effect to the faucet? If not, why wouldn't it be possible? If so, are there any good references for how to go about doing this?

Thanks,
Eric

 

[d3mm]

It is definitely possible. The field of engineering is called "Active Noise Cancellation". You can buy systems that do this already.

As for how to do it, I would recommend looking at sound processing software libraries, for example "pyo" for Python. You will of course need a microphone.

 

[sophiecentaur]

The most effective noise cancellation will be when the sensing mike and the loudspeaker are as close to the noise source as possible. The loudspeaker needs to produce sounds that are exactly opposite in phase and of equal amplitude to the noise you are trying to eliminate by destructive interference. Your 'system' needs to be able to identify the unwanted signal so the mike must be much closer to the noise source than the wanted source. Earphones are the easiest to treat, for this reason because you 'know' exactly what sound you want to hear - it has an electronic source. The region where you get cancellation gets tighter, the further apart you site the speaker and the mike. The snag about your particular application is the sound you are trying to eliminate because it's largely HF (hiss) and the wavelength will be only a few tens of cm and cancellation region will be very local.
If you could be sure that the noise source is located in a small area (i.e. noise not getting to you via the extended pipe system but just from the faucet) then you could possibly put a microphone and Speaker right next to the faucet. You could then get an small range of angles from the noise source where the cancellation is significant.
Afaik, most of the noise reduction systems that are available rely on the noise being easily identifiable - like, for instance, when you put a mike in the engine compartment and reduce the noise in the cab - but then the frequencies are low to mid range (much easier).
You could have some fun trying to make it work - you could finally 'help it along' by moving your head about to get the best result (crick your neck haha). It's great that processors work so fast these days that real time DSP is very achievable.

 

[d3mm]

Running water is low frequency. Tap hiss is not.

I have seen a marine installation of this that uses active cancellation for HVAC and engine noise (all low frequency) with padding (foam etc in the deck, overheads and ceilings) for high frequency.

 

[sophiecentaur]

Running water is low frequency. Tap hiss is not.

I have seen a marine installation of this that uses active cancellation for HVAC and engine noise (all low frequency) with padding (foam etc in the deck, overheads and ceilings) for high frequency.

It may be that the unwanted noise is mid / low frequency; we'd need to be told.
You re-iterate the point that physical sound insulation is also required when the source is 'mechanical'. This could be something that the OP needs to take into account.

I suppose it may be possible to use a directional microphone arrangement to select the source of the noise but getting the phases correct, in view of multiple reflections in the room, could be a limiting factor.
I imagine that a 'sweet spot' could be generated though, at the expense of other parts of the room.