Skip to content
GitLab
Commit e98a0034 authored by Mert Burkay Çöteli's avatar Mert Burkay Çöteli
Browse files

memory allocation problems are solved and ReadMe updates for the current AIMs

parent c9783dab
Hide whitespace changes
Inline Side-by-side
AnalysisTransform/README.md
View file @ e98a0034
...@@ -36,17 +36,14 @@ This README would normally document whatever steps are necessary to get your app ...@@ -36,17 +36,14 @@ This README would normally document whatever steps are necessary to get your app
// The output will be Interleaved Nofchannelsx10x1024 real and imaginary parts. // The output will be Interleaved Nofchannelsx10x1024 real and imaginary parts.
// Input and output data formats can be followed from MPAI-CAE v1.3 // Input and output data formats can be followed from MPAI-CAE v1.3
analysistransform* m_fft = new analysistransform(); analysistransform* m_fft = new analysistransform(2048, 20480,1024,2);
int NofChannels = 32;
float* signal; // Data acquired from each microphone channel float* signal; // Data acquired from each microphone channel
float* realpart; // Real part to be malloc float* realpart; // Real part to be malloc
float* imagpart; // Imaginary part to be malloc float* imagpart; // Imaginary part to be malloc
for (int channel = 0; channel < NofChannels; channel++) m_fft->FFT(signal, realpart, imagpart);
{
m_fft->STFT(signal, realpart+channel, imagpart+channel)
}
### Who do I talk to? ### ### Who do I talk to? ###
* Repo owner or admin * Repo owner or admin
......
AnalysisTransform/analysistransform.cpp
View file @ e98a0034
...@@ -9,6 +9,11 @@ analysistransform::analysistransform() ...@@ -9,6 +9,11 @@ analysistransform::analysistransform()
this->windowsize = 2048; this->windowsize = 2048;
this->overlap = 1024; this->overlap = 1024;
this->nofmics = 32; this->nofmics = 32;
// Create a hamming window of appropriate length
window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
} }
// Create a constructor for analysis transform // Create a constructor for analysis transform
...@@ -18,6 +23,11 @@ analysistransform::analysistransform(int windowsize, int signallength, int hopsi ...@@ -18,6 +23,11 @@ analysistransform::analysistransform(int windowsize, int signallength, int hopsi
this->windowsize = windowsize; this->windowsize = windowsize;
this->overlap = hopsize; this->overlap = hopsize;
this->nofmics = nofmics; this->nofmics = nofmics;
// Create a hamming window of appropriate length
window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
} }
// Create a hamming window of windowLength samples in buffer // Create a hamming window of windowLength samples in buffer
...@@ -25,7 +35,7 @@ void analysistransform::hamming(int windowLength, double* buffer) { ...@@ -25,7 +35,7 @@ void analysistransform::hamming(int windowLength, double* buffer) {
for (int i = 0; i < windowLength; i++) { for (int i = 0; i < windowLength; i++) {
buffer[i] = 0.54 - (0.46 * cos(2 * PI * (i / ((windowLength - 1) * 1.0)))); buffer[i] = 0.54 - (0.46 * cos(2 * PI * (i / (((double)windowLength - 1.0) * 1.0))));
} }
} }
...@@ -49,13 +59,6 @@ void* analysistransform::FFT(double* signal, double* realpart, double* imagpart) ...@@ -49,13 +59,6 @@ void* analysistransform::FFT(double* signal, double* realpart, double* imagpart)
plan_forward = fftw_plan_dft_r2c_1d(this->windowsize, data, fft_result, FFTW_ESTIMATE); plan_forward = fftw_plan_dft_r2c_1d(this->windowsize, data, fft_result, FFTW_ESTIMATE);
// Create a hamming window of appropriate length
double* window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
int readIndex; int readIndex;
// Should we stop reading in chunks? // Should we stop reading in chunks?
...@@ -124,7 +127,6 @@ void* analysistransform::FFT(double* signal, double* realpart, double* imagpart) ...@@ -124,7 +127,6 @@ void* analysistransform::FFT(double* signal, double* realpart, double* imagpart)
fftw_free(data); fftw_free(data);
fftw_free(fft_result); fftw_free(fft_result);
free(window);
return 0; return 0;
......
AnalysisTransform/analysistransform.h
View file @ e98a0034
...@@ -21,6 +21,7 @@ private: ...@@ -21,6 +21,7 @@ private:
int windowsize; int windowsize;
int overlap; int overlap;
int nofmics; int nofmics;
double* window;
}; };
......
README.md
View file @ e98a0034
...@@ -5,14 +5,14 @@ This README would normally document whatever steps are necessary to get your app ...@@ -5,14 +5,14 @@ This README would normally document whatever steps are necessary to get your app
### What is this repository for? ### ### What is this repository for? ###
* Quick summary * Quick summary
This repository comprises 4 AIM source codes (Analysis Transform, Sound Field Description, Synthesis Transform, and Packager) and 2 AIM executable (Speech Detection and Separation, Noise Cancellation) files This repository comprises 4 EAE AIM source codes (Analysis Transform, Sound Field Description, Synthesis Transform, and Packager) and 2 AIM executable (Speech Detection and Separation, Noise Cancellation) files
* Version * Version
v0.1 v0.1
### How do I get set up? ### ### How do I get set up? ###
* Summary of set up * Summary of set up
Each AIM's installation is defined in their ReadMe files Each AIM's installation is described in their ReadMe files
* Configuration * Configuration
* Dependencies * Dependencies
* Database configuration * Database configuration
......
SoundFieldDescription/soundfielddescription.cpp
View file @ e98a0034
...@@ -56,6 +56,8 @@ void* soundfielddescription::calcsphericalharmonics(double* pressure_real_in, do ...@@ -56,6 +56,8 @@ void* soundfielddescription::calcsphericalharmonics(double* pressure_real_in, do
} }
} }
} }
// Free the allocated space for spherical harmonics
free(sph_harmonics);
} }
return 0; return 0;
} }
\ No newline at end of file
SynthesisTransform/README.md 0 → 100644
View file @ e98a0034
# README #
This README would normally document whatever steps are necessary to get your application up and running.
### What is this repository for? ###
* Quick summary
Synthesis tranform is an AIM used for transferring the time-frequency domain multichannel signals into the time domain. This file introduces the setup for standalone application.
* Version
01.01.220708
* [Technical specifications MPAI-CAE v1.3](https://mpai.community/standards/resources/)
### How do I get set up? ###
* Summary of set up
The development language is C++. C++11 or its newer versions support compiling the code. See the list of C++ compilers. (https://en.wikipedia.org/wiki/List_of_compilers#C++_compilers)
This software uses a third-party fftw-3.3.9 that can be also downloaded from the link. (https://fftw.org/)
* Configuration
Usage of a new class
* Dependencies
include : \include
libs :
#if DEBUG
\libs\Windows_x64\Debug;\libs\Windows_x86\Debug;\libs\Linux\Debug
#else
\libs\Windows_x64\Release;\libs\Windows_x86\Release;\libs\Linux\Release
#endif
* Database configuration
This software doesnot use a database.
* How to run tests
[Conformance Testing Document MPAI-CAE](https://mpai.community/standards/resources/)
* Deployment instructions
***********Standalone usage of this AIM (code script)
// Input and output data formats can be followed from MPAI-CAE v1.3
synthesistransform* m_ifft = new synthesistransform();
float* signal; // Interleaved data output
float* realpart; // Real part to be malloc
float* imagpart; // Imaginary part to be malloc
m_ifft.IFFT(signal, realpart, imagpart);
### Who do I talk to? ###
* Repo owner or admin
Mert Burkay Çöteli , mbcoteli@aselsan.com.tr, Aselsan
* Other community or team contact
MPAI secreteriat , secretariat@mpai.community , MPAI
\ No newline at end of file
SynthesisTransform/synthesistransform.cpp
View file @ e98a0034
...@@ -3,22 +3,30 @@ ...@@ -3,22 +3,30 @@
// Create a constructor for synthesis transform // Create a constructor for analysis transform
synthesistransform::synthesistransform() synthesistransform::synthesistransform()
{ {
this->signallength = 20480; this->signallength = 20480;
this->windowsize = 2048; this->windowsize = 2048;
this->overlap = 1024; this->overlap = 1024;
this->nofmics = 32; this->nofmics = 32;
window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
} }
// Create a constructor for synthesis transform // Create a constructor for analysis transform
synthesistransform::synthesistransform(int windowsize, int signallength, int hopsize, int nofmics) synthesistransform::synthesistransform(int windowsize, int signallength, int hopsize, int nofmics)
{ {
this->signallength = signallength; this->signallength = signallength;
this->windowsize = windowsize; this->windowsize = windowsize;
this->overlap = hopsize; this->overlap = hopsize;
this->nofmics = nofmics; this->nofmics = nofmics;
window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
} }
// Create a hamming window of windowLength samples in buffer // Create a hamming window of windowLength samples in buffer
...@@ -26,11 +34,11 @@ void synthesistransform::hamming(int windowLength, double* buffer) { ...@@ -26,11 +34,11 @@ void synthesistransform::hamming(int windowLength, double* buffer) {
for (int i = 0; i < windowLength; i++) { for (int i = 0; i < windowLength; i++) {
buffer[i] = 0.54 - (0.46 * cos(2 * PI * (i / ((windowLength - 1) * 1.0)))); buffer[i] = 0.54 - (0.46 * cos(2 * PI * (i / (((double)windowLength - 1.0) * 1.0))));
} }
} }
// Function for the inverse fast fourier transform
void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imagpart) { void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imagpart) {
...@@ -38,17 +46,11 @@ void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imag ...@@ -38,17 +46,11 @@ void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imag
fftw_plan plan_backward; fftw_plan plan_backward;
int i; int i;
data = (fftw_complex*)fftw_malloc(sizeof(fftw_complex) * this->windowsize); data = (fftw_complex*)fftw_malloc(sizeof(fftw_complex) * this->windowsize);
ifft_output = (fftw_complex*)fftw_malloc(sizeof(fftw_complex) * this->windowsize); ifft_output = (fftw_complex*)fftw_malloc(sizeof(fftw_complex) * this->windowsize);
int outputcounter = 0; int outputcounter = 0;
double* window = (double*)malloc(sizeof(double) * this->windowsize);
hamming(this->windowsize, window);
plan_backward = fftw_plan_dft_1d(this->windowsize, data, ifft_output, FFTW_BACKWARD, FFTW_ESTIMATE); plan_backward = fftw_plan_dft_1d(this->windowsize, data, ifft_output, FFTW_BACKWARD, FFTW_ESTIMATE);
int readIndex; int readIndex;
...@@ -88,7 +90,10 @@ void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imag ...@@ -88,7 +90,10 @@ void* synthesistransform::IFFT(double* istft_out, double* realpart, double* imag
// Copy the first (windowSize/2 + 1) data points into your spectrogram. // Copy the first (windowSize/2 + 1) data points into your spectrogram.
// We do this because the FFT output is mirrored about the nyquist // We do this because the FFT output is mirrored about the nyquist
// frequency, so the second half of the data is redundant. // frequency, so the second half of the data is redundant. This is how
// Matlab's spectrogram routine works.
//outputcounter = chunkPosition * this->nofmics + channelindex;
for (i = 0; i < this->windowsize/2; i++) for (i = 0; i < this->windowsize/2; i++)
{ {
......
SynthesisTransform/synthesistransform.h
View file @ e98a0034
...@@ -21,6 +21,7 @@ private: ...@@ -21,6 +21,7 @@ private:
int windowsize; int windowsize;
int overlap; int overlap;
int nofmics; int nofmics;
double* window;
}; };
mainClass.cpp
View file @ e98a0034
#include "analysistransform.h" #include "analysistransform.h"
#include "synthesistransform.h" #include "synthesistransform.h"
#include "soundfielddescription.h" #include "soundfielddescription.h"
#include "packager.h"
int main() int main()
{ {
...@@ -40,25 +40,81 @@ int main() ...@@ -40,25 +40,81 @@ int main()
printf("{%f}", average_error / bloksize); printf("{%f}", average_error / bloksize);
soundfielddescription sfd = soundfielddescription();
char audioscenegeometry[] = R"( char audioscenegeometry[] = R"(
{ {
"$schema": "http://json-schema.org/draft-07/schema#", "$schema": "http://json-schema.org/draft-07/schema#",
"title" : "Audio Scene Geometry", "title" : "Audio Scene Geometry",
"BlockIndex": , "BlockIndex": 1,
"BlockStart" : , "BlockStart" : 1659533438878,
"BlockEnd" : , "BlockEnd" : 1659533438898,
"SpeechCount" : 1, "SpeechCount" : 1,
"SourceDetectionMask" : [], "SourceDetectionMask" : [1,1,1],
"SpeechList": [ "SpeechList": [
{"SpeechID" : 0.93358, "ChannelID" : 0.0, "AzimuthDirection" : 0.35836, "ElevationDirection" : 0.0, "DistanceFlag" : 0.93358, "Distance" : 0.0}, {"SpeechID" : "09859d16-3c73-4bb0-9c74-91b451e34925", "ChannelID" : 0.0, "AzimuthDirection" : 0.35836, "ElevationDirection" : 0.0, "DistanceFlag" : 0.93358, "Distance" : 0.0}
]
], }
}
)"; )";
char microphonearraygeometry[] = R"(
{
"$schema": "http://json-schema.org/draft-07/schema#",
"title": "Microphone Array Geometry",
"MicrophoneArrayType": 0,
"MicrophoneArrayScat": 0,
"MicrophoneArrayFilterURI": "filter.file",
"SamplingRate": 4,
"SampleType": 2,
"BlockSize": 5,
"NumberofMicrophones": 32,
"MicrophoneList" : [
{"xCoord" : 0.93358,"yCoord" : 0.0,"zCoord" : 0.35836,"directivity" : 0.0,"micxLookCoord" : 0.93358,"micyLookCoord" : 0.0,"miczLookCoord" : 0.35836},
{"xCoord" : 0.84804,"yCoord" : 0.52991,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : 0.84804,"micyLookCoord" : 0.52991,"miczLookCoord" : 0.0},
{"xCoord" : 0.93358,"yCoord" : 0.0,"zCoord" : -0.35836,"directivity" : 0.0,"micxLookCoord" : 0.93358,"micyLookCoord" : 0.0,"miczLookCoord" : -0.35836},
{"xCoord" : 0.84804,"yCoord" : -0.52991,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : 0.84804,"micyLookCoord" : -0.52991,"miczLookCoord" : 0.0},
{"xCoord" : 0.52991,"yCoord" : 0.0,"zCoord" : 0.84804,"directivity" : 0.0,"micxLookCoord" : 0.52991,"micyLookCoord" : 0.0,"miczLookCoord" : 0.84804},
{"xCoord" : 0.57922,"yCoord" : 0.57922,"zCoord" : 0.57357,"directivity" : 0.0,"micxLookCoord" : 0.57922,"micyLookCoord" : 0.57922,"miczLookCoord" : 0.57357},
{"xCoord" : 0.35836,"yCoord" : 0.93358,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : 0.35836,"micyLookCoord" : 0.93358,"miczLookCoord" : 0.0},
{"xCoord" : 0.57922,"yCoord" : 0.57922,"zCoord" : -0.57357,"directivity" : 0.0,"micxLookCoord" : 0.57922,"micyLookCoord" : 0.57922,"miczLookCoord" : -0.57357},
{"xCoord" : 0.52991,"yCoord" : 0.0,"zCoord" : -0.84804,"directivity" : 0.0,"micxLookCoord" : 0.52991,"micyLookCoord" : 0.0,"miczLookCoord" : -0.84804},
{"xCoord" : 0.57922,"yCoord" : -0.57922,"zCoord" : -0.57357,"directivity" : 0.0,"micxLookCoord" : 0.57922,"micyLookCoord" : -0.57922,"miczLookCoord" : -0.57357},
{"xCoord" : 0.35836,"yCoord" : -0.93358,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : 0.35836,"micyLookCoord" : -0.93358,"miczLookCoord" : 0.0},
{"xCoord" : 0.57922,"yCoord" : -0.57922,"zCoord" : 0.57357,"directivity" : 0.0,"micxLookCoord" : 0.57922,"micyLookCoord" : -0.57922,"miczLookCoord" : 0.57357},
{"xCoord" : -0.00625,"yCoord" : 0.35831,"zCoord" : 0.93358,"directivity" : 0.0,"micxLookCoord" : -0.00625,"micyLookCoord" : 0.35831,"miczLookCoord" : 0.93358},
{"xCoord" : 0.0,"yCoord" : 0.84804,"zCoord" : 0.52991,"directivity" : 0.0,"micxLookCoord" : 0.0,"micyLookCoord" : 0.84804,"miczLookCoord" : 0.52991},
{"xCoord" : 0.0,"yCoord" : 0.85716,"zCoord" : -0.51503,"directivity" : 0.0,"micxLookCoord" : 0.0,"micyLookCoord" : 0.85716,"miczLookCoord" : -0.51503},
{"xCoord" : 0.006254,"yCoord" : 0.35831,"zCoord" : -0.93358,"directivity" : 0.0,"micxLookCoord" : 0.006254,"micyLookCoord" : 0.35831,"miczLookCoord" : -0.93358},
{"xCoord" : -0.93358,"yCoord" : 0.0,"zCoord" : 0.35836,"directivity" : 0.0,"micxLookCoord" : -0.93358,"micyLookCoord" : 0.0,"miczLookCoord" : 0.35836},
{"xCoord" : -0.84804,"yCoord" : -0.52991,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : -0.84804,"micyLookCoord" : -0.52991,"miczLookCoord" : 0.0},
{"xCoord" : -0.93358,"yCoord" : 0.0,"zCoord" : -0.35836,"directivity" : 0.0,"micxLookCoord" : -0.93358,"micyLookCoord" : 0.0,"miczLookCoord" : -0.35836},
{"xCoord" : -0.84804,"yCoord" : 0.52991,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : -0.84804,"micyLookCoord" : 0.52991,"miczLookCoord" : 0.0},
{"xCoord" : -0.52991,"yCoord" : 0.0,"zCoord" : 0.84804,"directivity" : 0.0,"micxLookCoord" : -0.52991,"micyLookCoord" : 0.0,"miczLookCoord" : 0.84804},
{"xCoord" : -0.57922,"yCoord" : -0.57922,"zCoord" : 0.57357,"directivity" : 0.0,"micxLookCoord" : -0.57922,"micyLookCoord" : -0.57922,"miczLookCoord" : 0.57357},
{"xCoord" : -0.35836,"yCoord" : -0.93358,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : -0.35836,"micyLookCoord" : -0.93358,"miczLookCoord" : 0.0},
{"xCoord" : -0.57922,"yCoord" : -0.57922,"zCoord" : -0.57357,"directivity" : 0.0,"micxLookCoord" : -0.57922,"micyLookCoord" : -0.57922,"miczLookCoord" : -0.57357},
{"xCoord" : -0.52991,"yCoord" : 0.0,"zCoord" : -0.84804,"directivity" : 0.0,"micxLookCoord" : -0.52991,"micyLookCoord" : 0.0,"miczLookCoord" : -0.84804},
{"xCoord" : -0.57922,"yCoord" : 0.57922,"zCoord" : -0.57357,"directivity" : 0.0,"micxLookCoord" : -0.57922,"micyLookCoord" : 0.57922,"miczLookCoord" : -0.57357},
{"xCoord" : -0.35836,"yCoord" : 0.93358,"zCoord" : 0.0,"directivity" : 0.0,"micxLookCoord" : -0.35836,"micyLookCoord" : 0.93358,"miczLookCoord" : 0.0},
{"xCoord" : -0.57922,"yCoord" : 0.57922,"zCoord" : 0.57357,"directivity" : 0.0,"micxLookCoord" : -0.57922,"micyLookCoord" : 0.57922,"miczLookCoord" : 0.57357},
{"xCoord" : -0.00625,"yCoord" : -0.35831,"zCoord" : 0.93358,"directivity" : 0.0,"micxLookCoord" : -0.00625,"micyLookCoord" : -0.35831,"miczLookCoord" : 0.93358},
{"xCoord" : 0.0,"yCoord" : -0.84804,"zCoord" : 0.52991,"directivity" : 0.0,"micxLookCoord" : 0.0,"micyLookCoord" : -0.84804,"miczLookCoord" : 0.52991},
{"xCoord" : 0.0,"yCoord" : -0.84804,"zCoord" : -0.52991,"directivity" : 0.0,"micxLookCoord" : 0.0,"micyLookCoord" : -0.84804,"miczLookCoord" : -0.52991},
{"xCoord" : 0.00625,"yCoord" : -0.35831,"zCoord" : -0.93358,"directivity" : 0.0,"micxLookCoord" : 0.00625,"micyLookCoord" : -0.35831,"miczLookCoord" : -0.93358}
],
"MicrophoneArrayLookCoord" : {"xLookCoord" : 0.0, "yLookCoord" : 0.0, "zLookCoord" : 0.0}
}
)";
soundfielddescription sfd = soundfielddescription(microphonearraygeometry);
int msgsize = 0;
packager m_packager;
unsigned char* result = m_packager.createdata(audioscenegeometry, microphonearraygeometry, signal_out, &msgsize);
return 0; return 0;
} }
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment