|
Dear, I am currently do some testing in sound localization, but now i got one problem.........still can not be solve......:( The ALgorithm located on his book "DSP Applications using C and the TMS3206x DSK" "Page 268 Acoustic Direction Tracker". There are no problem for the simulation, when the input signal injected by function generators directly, but when the input signals coming from real world (microphones buit-in PCM3003)then the simulation is only can allocate/detect the angle from approximatly 45 degree to -45 degree...why? First i guess the problem may come from echo and noise....but the simulation can work in function generators signal.... why it can not work in real world signal?? Any idea??? Please... Same Algorithm, the equation given from the book....c=arcsin (a/b),......the time delay (phase/frequency)....is the frequency as input frequency and the phase is difference phase from FFT? Is't the setting jumper for PCM 3003 from the book has a mistake? So, what is the actuall setting and maximum sampling frequency for PCM3003? Do you have latest setting manual? Please... THANKS THANKS FOR YOUR TIME.... Sincerely Ehng3 |
|
|
Using PCM3003 to perform Sound localization
Started by ●April 29, 2004
Reply by ●April 30, 20042004-04-30
|
Hello Ehng3,
I do not know much about your subject, but i would expect that you need to
use a directional microphone to accomplish your purpose.
Regarding PCM3003 jumper settings - There are some mistakes. I
believe that they have been discussed several times in this group. You can
find the information in the archives.
mikedunnehng3 <e...@yahoo.com> wrote:
Dear, |
Reply by ●May 1, 20042004-05-01
|
Hi Mike and Ehng3 (nice name :) A directional mike seem to be a device to filter out acoustic waves coming from unwanted directions, the original question was about direction tracking. How many mikes are in the PCM3003 array? It looks like just 2. This imposes a number of constraints on direction tracking, including spatial (or directional) aliasing. I'd recommend Dudgeon and Mersereau text Multidimentional Digital Signal Processing, chpt 6. Rgds, Andrew > Date: Thu, 29 Apr 2004 19:43:35 -0700 (PDT) > From: Mike Dunn <> > Subject: Re: Using PCM3003 to perform Sound localization > > Hello Ehng3, > > I do not know much about your subject, but i would expect that you need > to use a directional microphone to accomplish your purpose. > > Regarding PCM3003 jumper settings - There are some mistakes. I believe > that they have been discussed several times in this group. You can find > the information in the archives. > > mikedunn > ehng3 <> wrote: > Dear, > > I am currently do some testing in sound localization, but now i got > one problem.........still can not be solve......:( > > The ALgorithm located on his book "DSP Applications using C and the > TMS3206x DSK" "Page 268 Acoustic Direction Tracker". There are no > problem for the simulation, when the input signal injected by > function generators directly, but when the input signals coming from > real world (microphones buit-in PCM3003)then the simulation is only > can allocate/detect the angle from approximatly 45 degree to -45 > degree...why? > > First i guess the problem may come from echo and noise....but the > simulation can work in function generators signal.... why it can not > work in real world signal?? Any idea??? Please... > > Same Algorithm, the equation given from the book....c=arcsin > (a/b),......the time delay (phase/frequency)....is the frequency as > input frequency and the phase is difference phase from FFT? > > Is't the setting jumper for PCM 3003 from the book has a mistake? > So, what is the actuall setting and maximum sampling frequency for > PCM3003? Do you have latest setting manual? Please... > THANKS THANKS FOR YOUR TIME.... > > Sincerely > Ehng3 |
Reply by ●May 3, 20042004-05-03
|
Ehng3 > I am currently do some testing in sound localization, but now i got > one problem.........still can not be solve......:( > > The ALgorithm located on his book "DSP Applications using C and the > TMS3206x DSK" "Page 268 Acoustic Direction Tracker". There are no > problem for the simulation, when the input signal injected by > function generators directly, but when the input signals coming from > real world (microphones buit-in PCM3003)then the simulation is only > can allocate/detect the angle from approximatly 45 degree to -45 > degree...why? > > First i guess the problem may come from echo and noise....but the > simulation can work in function generators signal.... why it can not > work in real world signal?? Any idea??? Please... > > Same Algorithm, the equation given from the book....c=arcsin > (a/b),......the time delay (phase/frequency)....is the frequency as > input frequency and the phase is difference phase from FFT? > > Is't the setting jumper for PCM 3003 from the book has a mistake? > So, what is the actuall setting and maximum sampling frequency for > PCM3003? Do you have latest setting manual? Please... I'm curious -- how is it going with this? Once you have PCM3003 daughtercard correctly configured (for 48 kHz Fs, I assume) and algorithm is correct, are you able to get past 45 deg with real signals? How far apart are your microphones? Depending on the sampling rate, you will need some minimum separation to obtain correct results... Also, are you connecting mics directly to PCM3003? Will that work, or does it require line-level input? -Jeff |
Reply by ●May 4, 20042004-05-04
|
Ng- > Thanks for your reply. > > I'm curious -- how is it going with this? Once you > > have PCM3003 daughtercard > > correctly configured (for 48 kHz Fs, I assume) and > > algorithm is correct, are you able > > to get past 45 deg with real signals? > > Ya i got the sampling approximately 48kHz and the > simulation given angle is around 45 to -45 degree. > what that simulation is only work for this range? is't > because of the "sin" function? how to solve that > problem so that i can get the angle or direction from > 180 to -180 degree. I think you mean 90 to -90 -- with only 2 mics you cannot resolve between front and rear. > > How far apart are your microphones? Depending on > > the sampling rate, you will need > > some minimum separation to obtain correct results... > > i'm just use the microphone from the PCM 3003, it will > cause the problem or not? Because i still can not get > the right angle in real time system. Actually i have > try using external pre-amp (two microphone also) to > test that simulation but still give me the problem. > Is't the distance between the microphone will cause > the problem to getting the angle or direction? > > Experiment set-up > The distance between source and microphone > approximately 60cm (more that one wavelength). and > distance between micphone approximately 6cm. > > > Also, are you connecting mics directly to PCM3003? > > Will that work, or does it > > require line-level input? > > Ya it was. but only work for 45 to -45 degree. I would record stereo signal at 90 deg and look at it with software to verify about 9 sample delay between L and R channels (for mic 6 cm apart). That will check your sampling rate and other assumptions. -Jeff > > -Jeff > --- Jeff Brower <> wrote: > > Ehng3 > > > > > I am currently do some testing in sound > > localization, but now i got > > > one problem.........still can not be solve......:( > > > > > > The ALgorithm located on his book "DSP > > Applications using C and the > > > TMS3206x DSK" "Page 268 Acoustic Direction > > Tracker". There are no > > > problem for the simulation, when the input signal > > injected by > > > function generators directly, but when the input > > signals coming from > > > real world (microphones buit-in PCM3003)then the > > simulation is only > > > can allocate/detect the angle from approximatly 45 > > degree to -45 > > > degree...why? > > > > > > First i guess the problem may come from echo and > > noise....but the > > > simulation can work in function generators > > signal.... why it can not > > > work in real world signal?? Any idea??? Please... > > > > > > Same Algorithm, the equation given from the > > book....c=arcsin > > > (a/b),......the time delay (phase/frequency)....is > > the frequency as > > > input frequency and the phase is difference phase > > from FFT? > > > > > > Is't the setting jumper for PCM 3003 from the book > > has a mistake? > > > So, what is the actuall setting and maximum > > sampling frequency for > > > PCM3003? Do you have latest setting manual? > > Please... > > > > I'm curious -- how is it going with this? Once you > > have PCM3003 daughtercard > > correctly configured (for 48 kHz Fs, I assume) and > > algorithm is correct, are you able > > to get past 45 deg with real signals? > > > > How far apart are your microphones? Depending on > > the sampling rate, you will need > > some minimum separation to obtain correct results... > > > > Also, are you connecting mics directly to PCM3003? > > Will that work, or does it > > require line-level input? > > > > -Jeff |
Reply by ●May 5, 20042004-05-05
|
EHNG3,
I am not sure where you are located or what resources that you have access
to... again, let me mention that I have only done some research on this subject
[some time ago].
If I was in your situation, I think that I would first try an
experiment. You can probably increase the directional sensitivity by
recessing your existing microphones 'in a tunnel'. Try
fabricating a small tube [paper, cardboard, or ??] around your microphones with
the inside diameter slightly larger [finished size] than the microphone.
Cover the inside surface with something to prevent reflections from the walls
[cloth or thin foam or ??]. I would start with a length of about 4 cm [I
am guessing] an rerun your previous tests and observe the results. I would
expect them to be improved from your original experiment. Of course there
may be better imformation availble "somewhere on the web".
Let us know how things work - even it it turns out that i am
incorrect.
Good luck,
mikedunn
ng enghsiung <e...@yahoo.com> wrote: Dear Mike, |
Reply by ●May 7, 20042004-05-07
|
Another thought here from a non-DSP guy. If directionality of the microphones is important to your project, you might try renting or borrowing some professional-quality microphones from a recording studio. They come in many directional flavors from omni-directional to shotgun microphones and parabolic microphones. They often require specialized power supplies and preamplifiers, but might put you on the right track. There are also many different combinations of stereo microphone techniques that enhance spacial localization from purely level methods to purely time delay methods and everything inbetween. A good place to start is the "Handbook for Sound Engineers" available from SAMS publishing. Of course, there are also microphones that work in more than two dimensions - including the Calrec Soundfeild microphone, but are rare and expensive. You can accomplish the same thing by using arrays of conventional microphones. Brian >From: Mike Dunn <> >To: ng enghsiung <> >CC: >Subject: Re: [c6x] Using PCM3003 to perform Sound localization >Date: Tue, 4 May 2004 20:24:22 -0700 (PDT) > >EHNG3, > >I am not sure where you are located or what resources that you have access >to... again, let me mention that I have only done some research on this >subject [some time ago]. > >If I was in your situation, I think that I would first try an experiment. >You can probably increase the directional sensitivity by recessing your >existing microphones 'in a tunnel'. Try fabricating a small tube [paper, >cardboard, or ??] around your microphones with the inside diameter slightly >larger [finished size] than the microphone. Cover the inside surface with >something to prevent reflections from the walls [cloth or thin foam or ??]. > I would start with a length of about 4 cm [I am guessing] an rerun your >previous tests and observe the results. I would expect them to be improved >from your original experiment. Of course there may be better imformation >availble "somewhere on the web". > >Let us know how things work - even it it turns out that i am incorrect. > >Good luck, >mikedunn > >ng enghsiung <> wrote: >Dear Mike, > >Thanks for you reply. You mean i can't use the >microphone directly from the PCM3003? I must build my >own microphone as array microphone? How about the >distance between of the microphone? It's this will >cause the problem? >Thanks > >Sinecely >EHNG3 >--- Mike Dunn wrote: > > Hello Ehng3, > > > > I do not know much about your subject, but i would > > expect that you need to use a directional microphone > > to accomplish your purpose. > > > > Regarding PCM3003 jumper settings - There are some > > mistakes. I believe that they have been discussed > > several times in this group. You can find the > > information in the archives. > > > > mikedunn > > > > > > ehng3 wrote: > > Dear, > > > > I am currently do some testing in sound > > localization, but now i got > > one problem.........still can not be solve......:( > > > > The ALgorithm located on his book "DSP Applications > > using C and the > > TMS3206x DSK" "Page 268 Acoustic Direction Tracker". > > There are no > > problem for the simulation, when the input signal > > injected by > > function generators directly, but when the input > > signals coming from > > real world (microphones buit-in PCM3003)then the > > simulation is only > > can allocate/detect the angle from approximatly 45 > > degree to -45 > > degree...why? > > > > First i guess the problem may come from echo and > > noise....but the > > simulation can work in function generators > > signal.... why it can not > > work in real world signal?? Any idea??? Please... > > > > Same Algorithm, the equation given from the > > book....c=arcsin > > (a/b),......the time delay (phase/frequency)....is > > the frequency as > > input frequency and the phase is difference phase > > from FFT? > > > > Is't the setting jumper for PCM 3003 from the book > > has a mistake? > > So, what is the actuall setting and maximum sampling > > frequency for > > PCM3003? Do you have latest setting manual? > > Please... > > > > > > THANKS THANKS FOR YOUR TIME.... > > > > Sincerely > > Ehng3 > > > > > > > > > > _____________________________________ > > Note: If you do a simple "reply" with your email > > client, only the author of this message will receive > > your answer. You need to do a "reply all" if you > > want your answer to be distributed to the entire > > group. > > > > _____________________________________ > > About this discussion group: > > > > To Join: Send an email to > > > > > > To Post: Send an email to > > > > To Leave: Send an email to > > > > > > Archives: http://www.yahoogroups.com/group/c6x > > > > Other Groups: http://www.dsprelated.com > > > > Yahoo! Groups Links > > > > > > > > > >__________________________________ > _________________________________________________________________ Mothers Day is May 9. Make it special with great ideas from the Mothers Day Guide! http://special.msn.com/network/04mothersday.armx |
Reply by ●May 13, 20042004-05-13
|
Ng- > Thanks for you reply, do you know how to measure the wavefront > (source to mics) or what is estimation measurement? Because i got one > problem in measuring the angle at the different distance (source to > mics) and the resulting of the angle are always different at the > different distance. It is have any algorithm can solve the problem? > > Is't when frequency changing the resulting angle with be change at > the same position? > > How to estimation error bounds in real time system? Not sure... you are asking algorithm / application-specific questions rather than DSP questions. But I might suggest you use a sweep rather than constant sine function. for a constant wave it will be hard to get delay accurately. Have the source sweep from dc to half your PCM3003 capability (approx 20 kHz), then perform a cross-correlation function between the mic inputs and you should be able to obtain a reasonable estimate of delay. -Jeff > --- In , Jeff Brower <jbrower@s...> wrote: > > Ng- > > > > > Thanks for your reply. > > > > I'm curious -- how is it going with this? Once you > > > > have PCM3003 daughtercard > > > > correctly configured (for 48 kHz Fs, I assume) and > > > > algorithm is correct, are you able > > > > to get past 45 deg with real signals? > > > > > > Ya i got the sampling approximately 48kHz and the > > > simulation given angle is around 45 to -45 degree. > > > what that simulation is only work for this range? is't > > > because of the "sin" function? how to solve that > > > problem so that i can get the angle or direction from > > > 180 to -180 degree. > > > > I think you mean 90 to -90 -- with only 2 mics you cannot resolve > between front and > > rear. > > > > > > How far apart are your microphones? Depending on > > > > the sampling rate, you will need > > > > some minimum separation to obtain correct results... > > > > > > i'm just use the microphone from the PCM 3003, it will > > > cause the problem or not? Because i still can not get > > > the right angle in real time system. Actually i have > > > try using external pre-amp (two microphone also) to > > > test that simulation but still give me the problem. > > > Is't the distance between the microphone will cause > > > the problem to getting the angle or direction? > > > > > > Experiment set-up > > > The distance between source and microphone > > > approximately 60cm (more that one wavelength). and > > > distance between micphone approximately 6cm. > > > > > > > Also, are you connecting mics directly to PCM3003? > > > > Will that work, or does it > > > > require line-level input? > > > > > > Ya it was. but only work for 45 to -45 degree. > > > > I would record stereo signal at 90 deg and look at it with software > to verify about 9 > > sample delay between L and R channels (for mic 6 cm apart). That > will check your > > sampling rate and other assumptions. > > > > -Jeff > > > > > > -Jeff > > > --- Jeff Brower <jbrower@s...> wrote: > > > > Ehng3 > > > > > > > > > I am currently do some testing in sound > > > > localization, but now i got > > > > > one problem.........still can not be solve......:( > > > > > > > > > > The ALgorithm located on his book "DSP > > > > Applications using C and the > > > > > TMS3206x DSK" "Page 268 Acoustic Direction > > > > Tracker". There are no > > > > > problem for the simulation, when the input signal > > > > injected by > > > > > function generators directly, but when the input > > > > signals coming from > > > > > real world (microphones buit-in PCM3003)then the > > > > simulation is only > > > > > can allocate/detect the angle from approximatly 45 > > > > degree to -45 > > > > > degree...why? > > > > > > > > > > First i guess the problem may come from echo and > > > > noise....but the > > > > > simulation can work in function generators > > > > signal.... why it can not > > > > > work in real world signal?? Any idea??? Please... > > > > > > > > > > Same Algorithm, the equation given from the > > > > book....c=arcsin > > > > > (a/b),......the time delay (phase/frequency)....is > > > > the frequency as > > > > > input frequency and the phase is difference phase > > > > from FFT? > > > > > > > > > > Is't the setting jumper for PCM 3003 from the book > > > > has a mistake? > > > > > So, what is the actuall setting and maximum > > > > sampling frequency for > > > > > PCM3003? Do you have latest setting manual? > > > > Please... > > > > > > > > I'm curious -- how is it going with this? Once you > > > > have PCM3003 daughtercard > > > > correctly configured (for 48 kHz Fs, I assume) and > > > > algorithm is correct, are you able > > > > to get past 45 deg with real signals? > > > > > > > > How far apart are your microphones? Depending on > > > > the sampling rate, you will need > > > > some minimum separation to obtain correct results... > > > > > > > > Also, are you connecting mics directly to PCM3003? > > > > Will that work, or does it > > > > require line-level input? > > > > > > > > -Jeff |
