Acoustic Beamforming Using a Rectangular Microphone Array

Using the Mathworks example:

http://www.mathworks.com/products/phased-array/examples.html?file=/products/demos/shipping/phased/microphonebeamformerdemo.html

I am reproducing this code using a rectangular 4x4 array rather than the linear array. I am in the header: Simulate the Received Signals

i am only using one audio signal x1 how do i use the step function?

i tried

temp=step(hCollector, x1,ang_dft+sqrt(noisePwr)*rand(rs,NSamPerFrame,[4
    4]));

The error I received is:

Warning: Input arguments must be scalar. 
Error using  + 
Matrix dimensions must agree.

Error in file (line 47)
temp=step(hCollector, x1,ang_dft+sqrt(noisePwr)*rand(rs,NSamPerFrame,[4
    4]));

how do i correctly execute this?

here is my code:

H = phased.OmnidirectionalMicrophoneElement('BackBaffled',true)
fc=[200 300 400]
ang = [0;0];
resp=step(H,fc,ang);
plotResponse(H,200,'RespCut','Az','Format','Polar')

ha = phased.URA([4 4],[0.00635 0.00635],'Element',H)

figure;
plotResponse(ha,200,340,'RespCut','3D','Format','Polar')

%---------------------

ang_dft = [-10,10]

fs= 8000

hCollector = phased.WidebandCollector('Sensor',ha,'PropagationSpeed',340,'SampleRate',fs,'ModulatedInput',false);
t_duration = 3;
t=0:1/fs:t_duration/fs;

rs= RandStream.create('mt19937ar','Seed',2008)
noisePwr = 1e-4 

%preallocate

NSamPerFrame =1000
NTSample = t_duration*fs
sigArray = zeros(NTSample,[4 4])
voice_dft= zeros(NTSample,1)

%set audio player

isAudioSupported = audiodevinfo(0)

if isAudioSupported
    hap = dsp.AudioPlayer('SampleRate',fs)
end

%simulate

for m=1:NSamPerFrame:NTSample
    sig_idx=m:m+NSamPerFrame-1
    x1=wavread('dft_voice_8kHz',[sig_idx(1) sig_idx(end)])
    temp=step(hCollector, x1,ang_dft+sqrt(noisePwr)*rand(rs,NSamPerFrame,[4 4]));
    if isAudioSupported
        step(hap,temp(:,3));
    end
    sigArray(sig_idx,:)=temp;
    voice_dft(sig_idx)=x1;
end

plot(t,sigArrary(:,3));

[Nick Yiw]

Hi, I am a student currently working on a project that involves distance speech recognition for controlling output. I really want to know what are the best recommendations of microphone arrays/chips that I can use which are compatible with MATLAB. Thank you.

Answer 1

[Honglei Chen]

The error is because you are adding noise to the angle. I believe it's a typo. That line should just reads like this

    temp=step(hCollector, x1,ang_dft)+sqrt(noisePwr)*rand(rs,NSamPerFrame,16);

Also, in the earlier part of the script, where you instantiate sigArray, it should be

    sigArray = zeros(NTSample,16);

HTH

[Osheen]

Hello Honglei, Thank you for your response. I am wondering is it possible use a frost beamformer on a signal that has been Time Delay beamformed? I am still using the following Mathworks example:

http://www.mathworks.com/products/phased-array/examples.html?file=/products/demos/shipping/phased/microphonebeamformerdemo.html

Below is my code and you can see I have made some adjustments based on the tutorial. This is error I received:

Error using phased.TimeDelayBeamformer/step
Too many input arguments; expected 1 (in
addition to the object handle), got 3.

Error in timefrost (line 83)
    temp =
    step(hbf,hbf2,step(hsig),ang_dft);
----------------------------------------------------------------

    H = phased.OmnidirectionalMicrophoneElement('BackBaffled',false)
          fc=[200 300 400]
          ang = [0;0];
          resp=step(H,fc,ang);
          plotResponse(H,200,'RespCut','Az','Format','Polar')

      ha = phased.URA([4 4],[0.00635 0.00635],'Element',H)
      c = 340;

      figure;
      plotResponse(ha,200,340,'RespCut','3D','Format','Polar')

      %---------------------

      ang_dft = [-30;0]
      ang_cleanspeech = [-10;10]
      ang_laughter = [20;0]

      fs= 8000

      hCollector = phased.WidebandCollector('Sensor',ha,'PropagationSpeed',340,'SampleRate',fs,'ModulatedInput',false);
      t_duration = 3;
      t=0:1/fs:t_duration-1/fs;

      rs= RandStream.create('mt19937ar','Seed',2008)
      noisePwr = 1e-4 

      %preallocate

      NSamPerFrame =1000
      NTSample = t_duration*fs
      sigArray = zeros(NTSample,16);
      voice_dft= zeros(NTSample,1);
      voice_cleanspeech = zeros(NTSample,1);
      voice_laugh = zeros(NTSample,1);

      %set audio player

      isAudioSupported = audiodevinfo(0);

      if isAudioSupported
          hap = dsp.AudioPlayer('SampleRate',fs)
      end

      %simulate

      for m=1:NSamPerFrame:NTSample
          sig_idx=m:m+NSamPerFrame-1
          x1 = wavread('dft_voice_8kHz',[sig_idx(1) sig_idx(end)]);
          x2 = wavread('cleanspeech_voice_8kHz',[sig_idx(1) sig_idx(end)]);
          x3 = 2*wavread('laughter_8kHz',[sig_idx(1) sig_idx(end)])

          temp=step(hCollector,[x1 x2 x3],[ang_dft ang_cleanspeech ang_laughter])+sqrt(noisePwr)*rand(rs,NSamPerFrame,16);
          if isAudioSupported
              step(hap,temp(:,1));
          end
          sigArray(sig_idx,:)=temp;
          voice_dft(sig_idx)=x1;
          voice_cleanspeech(sig_idx) = x2;
          voice_laugh(sig_idx) = x3;
      end

      figure;
      plot(t,sigArray(:,3));
      xlabel('Time (sec)'); ylabel('Amplitude (v)');
      title('Signal Received at Channel 3'); ylim([-3 3]);

      %Process with a time Delay Beamformer

      angSteer = ang_dft;
      hbf = phased.TimeDelayBeamformer('SensorArray',ha,'SampleRate',fs,'Direction',angSteer,'PropagationSpeed',c)
      hbf2 = phased.FrostBeamformer('SensorArray',ha,'SampleRate',fs,'PropagationSpeed',c,'FilterLength',20,'DirectionSource','Input port');

      hsig = dsp.SignalReader('Signal',sigArray,'SamplesPerFrame',NSamPerFrame);

      cbfOut = zeros(NTSample,1);

      for m= 1:NSamPerFrame:NTSample
          temp = step(hbf,hbf2,step(hsig),ang_dft);
          if isAudioSupported
              step(hap,temp);
          end
          cbfOut(m:m+NSamPerFrame-1,:) = temp;
      end

      figure
      plot(t,cbfOut);
      xlabel('Time (sec)'); ylabel('Amplitude (V)');
      title('Time Delay Beamformer Output');ylim([-3,3]);

[Honglei Chen]

Hi OSheen,

The first step of a Frost beamformer is time delay steering so there is no need to add a time delay beamformer in front of Frost beamformer.

As to the error you saw, it is because in your configuration, the steering angle for the time delay beamformer is specified in the property of 'Direction'. Therefore, the syntax for performing time delay beamforming is

y = step(hbf,x)

If you really want to add a Frost beamformer afterwards, what you need to do is passing the y into the Fromst beamformer, e.g.

y2 = step(hbf2,y,ang_dft)

HTH

[Osheen]

Thank you for your helpful response.

How would I apply the Frost beamformer on REAL data collected from a 4x4 microphone array? Also what is the best way to determine the FIR filter length of the beamformer?

Here is some code I have written so far to apply on real collected data. Please let me know if I am missing something

data = getdata(ai)  
hmic=phased.OmnidirectionalMicrophoneElement;
ha=phased.URA([4 4],[0.0635 0.0635],'Element',hmic);

      hsig=dsp.SignalReader('Signal',data,'SamplesPerFrame',NSampPerFrame);
      frost=phased.FrostBeamformer('SensorArray',ha,'SampleRate',Fs,...
         'PropagationSpeed',c,'FilterLength',100,'DirectionSource','Input port');

      frostout=zeros(NTSample,1);
      for m=1:NSampPerFrame:nt
         frostout(m:m+NSampPerFrame-1,:)=step(frost,step(hsig),angSteer);
      end

      soundsc(frostout,Fs,16)
      wavwrite(frostout,'Frostbeamformed real data');
      plot(t,frostout);
      xlabel('Time (sec)'); ylabel ('Amplitude (V)');
      title('Frost Beamformer Output');

Answer 2

[Zeynep Ertekin]

Hi,

I need a 2d or 3d sound source localization code with command load; can anyone please help me. Any help will be highly apprecited.