 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Natural Science Forum / Physics / Acoustics / July 2005source tracking problem
Hi, Is there a way to track a moving speaker using multiple microphones in a noisy environments. Do I have to assume the speaker is far away the microphones? I apprecited any information. Thanks a lot. When you say "speaker" do you mean a person saying random things or do you possibly mean a loudspeaker producing a sound with a predictable pattern? > Hi, > > Is there a way to track a moving speaker using multiple microphones in > a noisy environments. Do I have to assume the speaker is far away the > microphones? I apprecited any information. Thanks a lot. You need to describe the physical situation in more detail and exactly what you want to determine. When you say ,"track" do you want to know the position of the speaker in three dimensional space? Or are you more interested in a general position of the "speaker", such as the the "speaker" is located to right now? Is this a "speaker", such as a person giving a lecture in an auditorium? In general, you should be able to track the position of the source (speaker) using multiple receivers (microphones). The receivers should be located outside of the area of the source. The more receivers you use, the better you can locate/track. You should use a minimum of three receivers. If it is a constant signal with a constant amplitude, then you can triangulate the location based on TOF (time of flight) and realative strength of the received signal at the various microphones. If it is a dynamic signal, such as a person speaking, then it gets more complicated, but could still be done using the same principals. However, you will need to perform some sort of signal processing in order to know that you are tracking the right signal. Hi, Thanks a lot. I am going to use multiple microphones to track a human speaker. However the microphones might be very close to the person, such as cell phone case. Also, I don't have any prior knowledge about the position of the person, and just hope to find it using some algorithms. Is there any avaliable methods to do so? Thanks. > You need to describe the physical situation in more detail and exactly > what you want to determine. When you say ,"track" do you want to know [quoted text clipped - 14 lines] > However, you will need to perform some sort of signal processing in > order to know that you are tracking the right signal. Are there going to be many stationary microphones that the person will be walking by? I think this group did something like what you describe - http://www.bee.qut.edu.au/projects/saivt/speechaudiolab/ Andrew > Are there going to be many stationary microphones that the person will > be walking by? Hi, I think this group did something similar http://www.bee.qut.edu.au/projects/saivt/speechaudiolab/ Andrew > Hi, Thanks a lot. I am going to use multiple microphones to track a > human speaker. However the microphones might be very close to the > person, such as cell phone case. Also, I don't have any prior knowledge > about the position of the person, and just hope to find it using some > algorithms. Is there any avaliable methods to do so? Thanks. If there are two microphones on a horizontal line, and the task is to determine the azimuth of the talker, then the "zero crossings" technique will work. Here, the signal from a microphone is amplified to a very high level, clipping occurs, but such ac waves, no matter what frequency, always cross zero twice per cycle. A zero crossing detector (a common DSP algorithm) then signals that fact every time. A coincidence error detector is then applied, whose output of course is the time delay of one signal vs the other. When the talker is on the normal to the ,line between mics, this error is zero. If it is not zero, it's value is proportional the time delay and the microphone spacing. Now you can do one of two things: you can either compute the azimuth, or you can feed the error signal to a motor that drives a platform to position the microphone normal line at the talker. For voice tracking in a noisy environment, you have to select a narrow band of sound... perhaps the voice fundamental, perhaps the vowel area around 500 Hz, I'm not sure. That narrow band signal is then the signal to which the above mentioned DSP is applied. For multiple microphones, A more sophisticated matrix approach with regard to the subsequent handling of the error signals between many microphone pairs is required; invention left up to you. Angelo Campanella |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.