configurewhereisit.m

此程序使用声音来定位物体的位置

function configurewhereisit(fo)
%CONFIGUREWHEREISIT Configuration for WHEREISIT.
% CONFIGUREWHEREISIT will display a series of dialogs which will prompt 
% you to place a speaker at a specified location.  Once completed,
% the resulting data is displayed.    
%
% configurewhereisit uses a single tone at 500 Hz
% configurewhereisit(fo) uses a single tone at frequency fo Hz
% configurewhereisit(wavfilename) plays the contents of wavfilename instead
%
% See also WHEREISIT, MAKEASOUND, TRACKER.

% Loren Dean (original version)
% Scott Hirsch (this version)

filename='';        %Name of wav file for sound source (default: single tone)

if nargin==0
    fo = 500;
    
    
elseif isstr(fo) & exist(fo,'file');        %Specify file name
    filename=fo;
    
    %We still want to know the fundamental frequency.  We'll guess from spectral
    % analysis of the contents of the wav file
    
    [y,fs]=wavread(filename);
    [Pxx,F]=psd(y(:,1),min([length(y) 1024*4]),fs);
    [P_max,Max_ind] = max(Pxx);
    %watch out for big DC offset
    if Max_ind<4        %Any of the first few frequencies
        [P_max,Max_ind] = max(Pxx(4:end));
        Max_ind=Max_ind+3;
    end;
    fo = F(Max_ind);
end;

%Grab a handle to the Tracker window to find out which products we have
set(0,'ShowHiddenHandles','on');
tracker_fig = findobj('Name','Tracker');
tracker_handles = guidata(tracker_fig);
has = tracker_handles.has;
set(0,'ShowHiddenHandles','off');

%Create the figure
fh = figure('Name','Configure Acoustic Tracker', ...
    'NumberTitle','off');

h=msgbox(['We will now configure this experiment by placing the ' ...
        'speaker at two different locations.  You will be ' ...
        'prompted with three more dialogs.  Follow the ' ...
        'instructions given with each dialog.'], ...
    'Where is it? Configuration');
waitfor(h)

% Get the data when the speaker is placed next to the left microphone
[data(1,:),trace]=LRetrieveData( ...
    ['Please place the speaker next to the left microphone and ' ...
        'press OK when it is there.  A tone will sound for 1 second ' ...
        'after which you will be prompted with another dialog.'], ...
    'Left Microphone',fo,filename);

%Plot the data - top left of figure
time = linspace(0,1,length(trace));
subplot(4,3,1);
plot(time,trace(:,1));
ylabel('Left Mic');
title('Left Tone');
ax = axis;
subplot(4,3,4);
plot(time,trace(:,2));
ylabel('Right Mic');
xlabel('Time (s)');
axis(ax);

if data(1,1) < data(1,2)        %Mics are switched
    errordlg('Your mics are crossed.  Please switch the left and right mics.');
    %close(fh)
    return
end;


% Get the data when the speaker is placed next to the right microphone
[data(2,:),trace]=LRetrieveData( ...
    ['Please place the speaker next to the right microphone and ' ...
        'press OK when it is there.  A tone will sound for 1 second ' ...
        'after which you will be prompted with another dialog.'], ...
    'Right Microphone',fo,filename);

%Plot the data - top right
subplot(4,3,3);
plot(time,trace(:,1));
title('Right Tone');
axis(ax)
subplot(4,3,6);
plot(time,trace(:,2));
xlabel('Time (s)');
axis(ax);

% Get the data when the speaker is placed in the middle.  We'll use this
%  for mic calibration and for placement guess
[data(3,:), trace]=LRetrieveData( ...
    ['Please place the speaker mid-way between the two microphones.' ...
        'A tone will sound for 1 second ' ...
        'after which you will be prompted with another dialog.'], ...
    'Center',fo,filename);

%Plot the data - top center
subplot(4,3,2);
plot(time,trace(:,1));
title('Center Tone');
axis(ax)
subplot(4,3,5);
plot(time,trace(:,2));
xlabel('Time (s)');
axis(ax);


%Calibrate the microphones so that they have about the same respone
%There are a bunch of ways we could do this with the data collected
%The ideal would be to measure the response of both mics subjected
% to an identical sound.  
%If the third cal is done with the speaker truly in the middle, this is
% a good way to go.
if 0
    cal = data(3,:);        
end;
%We could also use the first two tones, assuming that the speaker is held 
% in about the same position relative to each mic for the two cals.
cal = data([1 5]);        %First mic, first test; second mic, second test

%Apply calibration factor to mic in second channel.
data(:,2) = data(:,2)*cal(1)/cal(2);


%Fit 1/r curve to data
x=[0 .5 1]';
y = data([1 3 2],:);

if has.cfit     %Has curve fitting toolbox
    
    warning off     %Hide warnings about not specifying the start point
    LeftMic = fit(x,y(:,1),'rat01');            %rat01: rational, 0 order num, 1st order den
    RightMic = fit(x,flipud(y(:,2)),'rat01');       %Use same 1/r, then reverse
    warning on
    
    %Plot results of curve fit
    subplot(2,2,3);
    plot(LeftMic,x,y(:,1));
    legend('data','1/r fit')
    ylabel('Left Mic');
    subplot(2,2,4);
    plot(RightMic,x,flipud(y(:,2)));
    legend('data','1/r fit',2)
    ylabel('Right Mic');
    set(gca,'XDir','reverse');
    
else        
    %People who don't have the curve fitting toolbox make my life more difficult!
    %Try to fit y = p1/(x+q1)
    %To use MATLAB's polyfit:
    % y = pl / (x + q1)
    % y_inv = 1/y = (x + q1) / p1
    % y_inv = (1/p1)*x^1 + (q1/p1)*x^0
    % y_inv = c1*x^1 + c0*x^0       %Fit this
    % then, p1 = 1/c1;
    %       q1 = co/c1;
    
    C_Left = polyfit(x,1./y(:,1),1);
    C_Right = polyfit(x,1./flipud(y(:,2)),1);
    p1_Left = 1/C_Left(1);
    q1_Left = C_Left(2) / C_Left(1);
    p1_Right = 1/C_Right(1);
    q1_Right = C_Right(2) / C_Right(1);
    
    %Build LeftMic, RightMic structures
    LeftMic.p1 = p1_Left;
    LeftMic.q1 = q1_Left;
    RightMic.p1 = p1_Right;
    RightMic.q1 = q1_Right;
    
    %Plot results of curve fit
    x_fit = linspace(0,1);
    
    LeftMic_fit = 1./polyval(C_Left,x_fit);
    RightMic_fit = 1./polyval(C_Right,x_fit);
    
    subplot(2,2,3);
    plot(x,y(:,1),'r.',x_fit,LeftMic_fit,'b');
    legend('data','1/r fit')
    ylabel('Left Mic');
    subplot(2,2,4);
    plot(x,flipud(y(:,2)),'r.',x_fit,RightMic_fit,'b');
    legend('data','1/r fit',2)
    ylabel('Right Mic');
    set(gca,'XDir','reverse');
    
end;



% Save the data
save whereisitconfiguration data fo LeftMic RightMic cal filename

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%% LRetrieveData %%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%
function [magnitude,data]=LRetrieveData(msgString,msgTitle,fo,filename)
% Function to make a sound, retrieve data for 1 second through two 
% microphones and then determine the magnitude of the incoming data.

% Display a message box to prompt the user where to place the speaker
h=msgbox(msgString,msgTitle);
waitfor(h)

% Start playing a sound
if exist(filename,'file')
    makeasound(filename);
else   
    makeasound(fo);
end;

% Create an input object with two channels and configure it to 
% run for 1 second
AI=analoginput('winsound');
addchannel(AI,1:2);


AI.SampleRate=44100; % Hz
AI.SamplesPerTrigger=AI.SampleRate*2;  % Run for 1 second

% Start the object and get the data out from its buffer
% The object will automatically stop when it has acquired the 
% number of samples specified by SamplesPerTrigger above.
start(AI)
data=getdata(AI);

% Stop playing the sound and delete the input object
makeasound stop
delete(AI)

% Determine the magnitude of the sound but only look at the data in the 
% middle of the dataset so that any transients are removed.
range=length(data)/4:3/4*length(data);
magnitude = localRMS(data(range,1:2));
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


function out=localRMS(in,dim);
%RMS             Compute rms value of a steady signal

[nr,nc]=size(in);
if nargin<1
    help(mfilename);
elseif nargin<2
    [junk,dim]=max([nr nc]);
end;

out=sqrt(mean(in.^2,dim));