Post by thread:[EE] Noise canceling

----- Original Message ----- From: "V G" <spam_OUTx.solarwind.xTakeThisOuTgmail.com> To: "PICLIST" <.....piclistKILLspam@spam@mit.edu> Sent: Monday, August 09, 2010 8:42 PM Subject: [EE] Noise canceling > Hey all, > > Does a product currently exist on the market that: > > Takes two or more microphone inputs and applies noise canceling to > them, then provides a noise-suppressed output? (Ex: by phase shifting > and adding the waves) It seems you are describing any of several noise-reduing headsets that are available from electronics retailers, office supply stores, and Wal-Mart, as well as others. > If not, how difficult would it be to make one? Perhaps with a dsPIC or > PIC32.

V G wrote: > Hey all, > > Does a product currently exist on the market that: > > Takes two or more microphone inputs and applies noise canceling to > them, then provides a noise-suppressed output? (Ex: by phase shifting > and adding the waves) > Doing it effectively requires a little more that phase shifting afaict, instead you have a system that dynamically adjusts a filter to try and match the magnitude and phase as closely as possible. The difference of the two signals (the "error signal" from the adaptive filter) then contains the stuff that was only picked up by one microphone (you.This is known as two input adaptive filtering. One of my lecturers at uni "Patrick Gaydecki" developed a general purpose DSP box that can do this. His device has a website at http://www.signalwizardsystems.com/ and it can be purchased from http://www.saelig.com/category/DA.htm . It seems a little on the expensive side for what it is though (I guess as with most such things the real value is in the firmware). He has also written a rather good book on DSP. www.amazon.co.uk/Foundations-Digital-Signal-Processing-Algorithms/dp/0852964315/ref=sr_1_1?ie=UTF8&s=books&qid=1281405047&sr=8-1 > If not, how difficult would it be to make one? Perhaps with a dsPIC or PIC32. > The hardware isn't that complex really, afaict the real effort is in developing the firmware but if you are just developing for one situation you can likely make it simpler than the stuff for a general purpose box

> > Does a product currently exist on the market that: > > > > Takes two or more microphone inputs and applies noise canceling to > > them, then provides a noise-suppressed output? (Ex: by phase shifting > > and adding the waves) > > It seems you are describing any of several noise-reduing headsets that are > available from electronics retailers, office supply stores, and Wal-Mart, as > well as others. That was my immediate thought as well. > > If not, how difficult would it be to make one? Perhaps with a dsPIC or > > PIC32. A dsPic would be potentially capable of doing this. Microchip does have an echo-cancelling library for it. -- Scanned by iCritical.

> > > Does a product currently exist on the market that: > > > Takes two or more microphone inputs and applies noise canceling to > > > them, then provides a noise-suppressed output? What they all said + The issue, which Peter alluded to an aspect of, is that you have to effectively model the space in which the sound is being generated and monitored. Your canceller does effectively do phase inversion as you suggest BUT the amplitude and phase of all frequency components depends critically on the overall environment. International telephone circuit echo cancellers do this but they have the advantage that their environment typically contains a boring old cable with some impedance bumps in it, rather than an auditorium of unknown size and shaped\e, many audio paths (direct plus reflections) all of which have differing frequency/amplitude/phase characteristics. Add people, chairs and tables, opening and shutting doors etc and you need very capable adaptation software. Simplistically this can probably be modelled as an IIR filter with variable dynamic taps - telephone cables can, but they are simpler as noted. The computer doesn't have to know about its environment - just be capable of handling the results. Some very very capable people have worked on all aspects of this general problem for a long while (eg I know a man who had worked out a canceller about 35 years ago but had to wait a decade or more while the available hardware caught up with his ideas). So, yes, it can be done and it is done, but you are in for some serious maths if you want to compete with the leaders.

> I was thinking about noise canceling at the source (microphone). Many > cell phones do this already with multiple microphone and they do it > well. Yes. There can be some very complex stuff inside consumer level equipment. That may not be so, but can be and quite invisibly. eg - I'm told that the best noise cancelling Bluetooth headset available will literally blank out an adjacent chainsaw. I'm told that you can here a chainsaw started up besides the wearer and then it just fades away to inaudibility. Apparently not so good for music though. (Guessing only: presumably the environmental adaptation tends to prevent a consistent channel model.) - Hearing aids used to be simple amplifiers with appropriate frequency shaping and some clipping. Nowadays the best in ear ones are tiny but reportedly have DSP capability with tailoring of response in multiple frequency bands and much more - and are remotely programmable. Some of the hearing aid techniques may be applicable to your requirement fwiw. Digital Hearing Aids: Current "State-of-the-Art" http://www.asha.org/public/hearing/treatment/digital_aid.htm SIGNAL PROCESSING TECHNIQUES FOR A DSP HEARING AID Multiple microphones and noise cancelling mentioned in this pper. Nothing very technical. http://brent.edwards.name/Papers/TAA11_7.PDF Digital Signal Processing and Hearing Aids 2002 - "everything has changed ... http://www.healthyhearing.com/articles/7719-digital-signal-processing-and A DSP implementation of a digital hearing aid with recruitment of loudness compensation and acoustic echo cancellation Sounds highly apposite - abstract only but you may have access via yoiur institution. http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=625579 Digital Hearing Aids - the Way of the Future http://www.digital-recordings.com/publ/pubaids.html Gennum Abandons Hearing-Aid Market With DSP Chip and Headset Spinoffs 2007. Interesting if not overly relevant :-) http://p6.hostingprod.com/@hearingmojo.com/blog-mt/blog-mt/2007/09/gennum_abandons_hearingaid_mar.htm

Gargoyle DSP +"echo canceliing" DSP +"noise cancelling" Starting here may be useful http://www.dsprelated.com/ Welcome to DSPRelated.com ! DSPRelated.com has been around since 1999 and has become over the years the most popular independent internet resource for DSP engineers around the world. You are invited to create an account with us (quick and free). Doing so will allow you to contribute to the the site by rating DSP books and DSP Internet resources, as well as participate to one or many DSP related discussion(s). _____________________________________ DSP noise cancelling project group 2006 page - may stilll be active http://www.dsprelated.com/showmessage/56897/1.php Worth copying here. Sounds like just what you want. _________________________ Hey everyone, I'm new to this forum. I am working with a group of Virginia Tech EE undergraduate students doing our senior project. We are designing a digital controller for noise canceling headphones. We have a pair of headphones equipped with an internal microphone. We've done initial testing of our system, and designed our controller. The transfer function is 4th order. Here it is: Zero/pole/gain: 0.15321 (z^2 + 1.248z + 0.7911) (z^2 + 0.2258z + 0.8837) ---------------------------------------------------------- (z^2 - 1.207z + 0.3842) (z^2 - 1.568z + 0.7318) Sampling time: 8e-0 We have a ADSP-2181 EZKit and are trying to implement this transfer function with the board. We are looking for any help in figuring out an easy way to program the board to achieve this transfer function. I know very little about this board, but have read some stuff on how to program specific filters. These sites specifically: www.ee.iastate.edu/~dickerson/classes/ee424/lab/lab1/LAB1A.html and this pdf http://www.ece.rutgers.edu/~orfanidi/ezkitl/man.pdf Any help is appreciated

National makes a chip LM1088. I've played with it, actually works well but not for our applications. ______________________________________ Listen Technologies Corporation Harrison Cooper Senior Product Development Engineer 801.233.8992 800.330.0891 North America http://www.listentech.com Donâ€™t miss a single sound. ListenTech Blog â— twitterÂ â—Â facebook â— Â YouTube Introducingâ€¦ ListenPointâ„¢ - Audio Structure for Rooms Made Easy Integrate all your sound sources in small to medium sized rooms to accelerate productivity up to 30%. ï please consider the environment before printing this e-mail note {Original Message removed}