odas
odas copied to clipboard
introlab
Cannot obtain SSL result on simulated audio signal
Hello, I'm testing odas sound source localization (ssl) with simulated audio signal.
First, I saved ssl result in json file (for example potential_ssl.json). Using python script, I read the json file and converted x,y,z and e values to elevation and azimuth degs using code written in odas_web/graph.js. When I plot the angels, I wasn't able to obtain ssl result on azimuth (expected azimuth deg=90 and -135).
Here I uploaded respeaker with polar coordinate.
I think there must be something wrong with ssl result. Would you please help me how to obtain azimuth angles accurately? Here is my configuration file: `# Configuration file for ReSpeaker circular sound card
version = "2.1";
Raw
raw: {
fS = 16000;
hopSize = 512;
nBits = 32;
nChannels = 7;
# Input with raw signal from file
interface: {
type = "file";
path = "input.raw";
};
}
Mapping
mapping: {
map: (1, 2, 3, 4, 5, 6, 7);
}
General
general: {
epsilon = 1E-20;
size:
{
hopSize = 512;
frameSize = 1024;
};
samplerate:
{
mu = 16000;
sigma2 = 0.01;
};
speedofsound:
{
mu = 343.0;
sigma2 = 25.0;
};
mics = (
# Microphone 1
{
mu = ( +0.0000, +0.0000, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 2
{
mu = ( -0.0160, +0.0277, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 3
{
mu = ( -0.0320, +0.0000, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 4
{
mu = ( -0.0160, -0.0277, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 5
{
mu = ( +0.0160, -0.0277, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 6
{
mu = ( +0.0320, +0.0000, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
},
# Microphone 7
{
mu = ( +0.0160, +0.0277, +0.0000 );
sigma2 = ( +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000, +0.000 );
direction = ( +0.000, +0.000, +1.000 );
angle = ( 80.0, 90.0 );
}
);
# Spatial filters to include only a range of direction if required
# (may be useful to remove false detections from the floor, or
# limit the space search to a restricted region)
spatialfilters = (
{
direction = ( +0.000, +0.000, +1.000 );
angle = (80.0, 90.0);
}
);
nThetas = 181;
gainMin = 0.25;
};
Stationnary noise estimation
sne: {
b = 3;
alphaS = 0.1;
L = 150;
delta = 3.0;
alphaD = 0.1;
}
Sound Source Localization
ssl: {
nPots = 2;
nMatches = 10;
probMin = 0.5;
nRefinedLevels = 1;
interpRate = 4;
# Number of scans: level is the resolution of the sphere
# and delta is the size of the maximum sliding window
# (delta = -1 means the size is automatically computed)
scans = (
{ level = 2; delta = -1; },
{ level = 4; delta = -1; }
);
# Output to export potential sources
potential: {
format = "json";
interface: {
type = "file";
path = "potential_ssl.json";
};
};
};
Sound Source Tracking
sst: {
# Mode is either "kalman" or "particle"
mode = "particle";
# Add is either "static" or "dynamic"
add = "dynamic";
# Parameters used by both the Kalman and particle filter
active = (
{ weight = 1.0; mu = 0.3; sigma2 = 0.0025 }
);
inactive = (
{ weight = 1.0; mu = 0.15; sigma2 = 0.0025 }
);
sigmaR2_prob = 0.0025;
sigmaR2_active = 0.0225;
sigmaR2_target = 0.0025;
Pfalse = 0.1;
Pnew = 0.1;
Ptrack = 0.8;
theta_new = 0.9;
N_prob = 5;
theta_prob = 0.8;
N_inactive = ( 150, 200, 250, 250 );
theta_inactive = 0.9;
# Parameters used by the Kalman filter only
kalman: {
sigmaQ = 0.001;
};
# Parameters used by the particle filter only
particle: {
nParticles = 1000;
st_alpha = 2.0;
st_beta = 0.04;
st_ratio = 0.5;
ve_alpha = 0.05;
ve_beta = 0.2;
ve_ratio = 0.3;
ac_alpha = 0.5;
ac_beta = 0.2;
ac_ratio = 0.2;
Nmin = 0.7;
};
target: ();
# Output to export tracked sources
tracked: {
format = "json";
interface: {
type = "file";
path = "tracks.txt";
};
};
}
sss: {
# Mode is either "dds", "dgss" or "dmvdr"
mode_sep = "dds";
mode_pf = "ms";
gain_sep = 1.0;
gain_pf = 10.0;
dds: {
};
dgss: {
mu = 0.01;
lambda = 0.5;
};
dmvdr: {
};
ms: {
alphaPmin = 0.07;
eta = 0.5;
alphaZ = 0.8;
thetaWin = 0.3;
alphaWin = 0.3;
maxAbsenceProb = 0.9;
Gmin = 0.01;
winSizeLocal = 3;
winSizeGlobal = 23;
winSizeFrame = 256;
};
ss: {
Gmin = 0.01;
Gmid = 0.9;
Gslope = 10.0;
}
separated: {
fS = 16000;
hopSize = 128;
nBits = 32;
interface: {
type = "file";
path = "separated.raw";
};
};
postfiltered: {
fS = 16000;
hopSize = 128;
nBits = 32;
gain = 10.0;
interface: {
type = "file";
path = "postfiltered.raw";
};
};
}
classify: {
frameSize = 1024;
winSize = 3;
tauMin = 32;
tauMax = 200;
deltaTauMax = 7;
alpha = 0.3;
gamma = 0.05;
phiMin = 0.15;
r0 = 0.2;
category: {
format = "undefined";
interface: {
type = "blackhole";
}
}
} ` I uploaded input.raw and input.wav file (simulated in matlab with repeaker microphone setting and recorded 2 persons who are speaking from 90 and -135 azimuth degree, 7 channels were saved in wav file and the wav file was converted into raw file using sox)
Also, I attached corresponding azimuth angle result with input waveform.
(ssl ground truth for 2 sources)
(ssl result from odas)
Thank you in advance. Baynaa.
