Hi Roman,
With the method I mentioned, the key thing is using high frequencies.
In addition to make the whole thing practically inaudible, it's also
perhaps the only way to avoid interference from reflections and
background noise. If you want or need to use lower frequency range
then you should try, e.g., correlation techniques.
As for the external, I think you should be able implemented it with
pure Pd as well. I actually tried to make a Pd-version of it but I
never really got it working (to be honest, I'm really clumsy with Pd)
and I just ended up implementing it in C.
Here's a short description what the external does for the received
signal:
- high-pass the signal with the same cut-off frequency as used for
the sent impulse (~ 16-17kHz) 2) rectify the signal 3) low-pass filter the signal below ca. 4 kHz 4) find the maximum (for more accuracy use interpolation)
The original idea came from my professor Matti Karjalainen. So the
credits goes to him :) I attached a demo patch that shows how the
signals look like in the first three steps (it requires iem-libraries).
I just compiled a linux version of the external but I first have to
get audio working in my linux-box so I can test it before I'll send
it. And sure, I can send the source and the make file I used, as well.
Cheers, Miikka
On Feb 19, 2007, at 5:46 PM, Roman Haefeli wrote:
hi mikka
thank you very much for all that detailed information and sorry for my late reply. i didn't have too much time and i really wanted to have a closer look at this before answering.
i am on linux, so i couldn't test your external. but from what you describe in your mail, it follows a similar approach i tried as well, the only difference is that yours obviously works, whereas i've had
bad luck yet with my tests ;-). maybe i am wrong, but i believe that the method you describe could be implemented in plain pd without using externals (this is at least what i hope).am i right in thinking that the analysis part of your external also
high pass filters the incoming signal from the mic and the measures the
time, until a certain amplitude of the signal is triggered? i have many problems implement this nicely. the main problem is i think is, that i maybe should do the trigger mechanism adaptively, so that it works on changing levels. how is that done in your external?if you don't mind, it would be nice to have the source code of your external, so that i could at least try to compile it, though i am
not a programmer myself and maybe i fail without a ready-made makefile. but it's sure worth a try.thank you again for sharing all that information about this topic.
best roman
On Wed, 2007-02-14 at 16:51 +0200, Miikka Tikander wrote:
Hi,
I made some externals for binaural acoustic positioning for my work some time ago and it also included some externals for acoustic distance measurements. I was supposed to polish them up to be published on my website but I never got that far... Anyway, here's one simple patch with one of the distance measurement externals.
The external is an OSX version but I know I have a linux version somewhere as well. If needed I can try to find it and send it as well, or maybe I'll send the source so people can compile it themselves if interested. It should be compilable WIN and Linux as
well.As there's no documentation in the patch, here's a brief documentation for the patch and the external.
The DIST_HP~ external sends high-passed (> 17 kHz) impulses from the loudspeaker and then measures the fly-time of these impulses to two microphones (here input channels 1 and 2). The slider on the upper right hand corner (Impuse interval) sets the time interval (in 64 sample blocks) of the impulses, that is, how often the distance is measured.
The last two outlets of the DIST_HP gives out the distance of the microphones to the loudspeaker in meters. I put a mean and a median filter on there as well to smooth the data but you may use whatever is appropriate. The distance is computed in sample accuracy.
The 'free_distance_in_meters' -outlet transforms the impulse time interval to a physical distance. This, in principal, should be greater than the longest microphone-loudspeaker distance you will
use.The calibration is done by using the calibration_latency -number on the left. Place a microphone to a known distance from the loudspeaker and then tune the calibration_latency so that the measured distance is correct. Increasing the latency increases the measured distance. The measured distance cycles from 0 to the free_distance_in_meters. So, when increase the calibration_latency and the measured distance reaches the free_distance_in_meters it will go back to zero and continue from there.
The volume slider sets the output volume of the impulses. Note, that the impulses are high passed above 17 kHz so actually not everybody will here them! Be cautious with the volume. The whole idea of using higher frequency range was to make the impulses inaudible, and to avoid interference from room reflection and from other sounds as well. I have a version which uses full band impulses as well, but then the reflections start to worsen things..
You must use 44,1 kHz sampling frequency.
Hope, it works and is usable. Once I get little extra time, I'll try to get the rest of the stuff organized and documented so I can publish them here as well. Any feedback or criticism is always
welcome.-miikka
On Feb 14, 2007, at 2:18 PM, Roman Haefeli wrote:
hi all
i'd like to create a patch, that measures the distance between the speaker and the microphone. my first problem is, that i don't know which kind of signal would be suitable for that purpose. the simplest
is to use a dirac impulse. but on the receiver-side it's hard to separate unwanted artefacts from the impulse itself. also it would be nice, if it would be possible to detect the signal reliably at different amplitudes, since the amplitude heavily changes when moving the mic closer to
and farer from the speaker. i am looking for a method, that allows
higher precision than the usual block-size (which introduces a grid of ~0.5m @ 44100Hz, which is not very accurate). also should it be possible to detect the signal correctly, when there was reverberation added
from a room. i know, there are ultrasonic-bases devices to measure a
distance to a wall, but i don't have clue how they work, but it makes me
hope, that something similar is possible with pd. any hints are much appreciated.roman
___________________________________________________________ Der frühe Vogel fängt den Wurm. Hier gelangen Sie zum neuen Yahoo! Mail: http://mail.yahoo.de
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
___________________________________________________________ Telefonate ohne weitere Kosten vom PC zum PC: http:// messenger.yahoo.de
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Miikka Tikander 050-543 4491