https://www.sofaconventions.org/mediawiki/api.php?action=feedcontributions&feedformat=atom&user=TcarpentSofaconventions - User contributions [en]2026-05-14T05:18:31ZUser contributionsMediaWiki 1.43.6https://www.sofaconventions.org/mediawiki/index.php?title=Files&diff=1893Files2014-11-07T13:10:01Z<p>Tcarpent: </p>
<hr />
<div><br />
<br />
The [http://sofacoustics.org/data main SOFA repository] aims at collect the worldwide available HRTFs, BRIRs, DRIRs, and other SOFA-related data at a single place. It is just in the process of being created: Partial download, metadata access, and database search is not available (yet, we are investigating the possibility of using [http://www.opendap.org/ OPeNDAP] for SOFA repositories). Currently, the data can be accessed and downloaded as they are and the metadata are provided in the particular files.<br />
<br />
== General purpose [http://sofacoustics.org/data/database database]: ==<br />
<br />
Standard (in-the-ear canal) HRTFs of humans: <br />
* [http://sofacoustics.org/data/database/ari ARI]: HRTFs from the [http://www.kfs.oeaw.ac.at/hrtf ARI database]. In-the-ear HRTFs and DTFs for over 100 listeners.<br />
** hrtf, dtf: HRTFs and DTFs, respectively, equalized between 300 Hz and 18 kHz<br />
** hrtf b, dtf b: HRTFs and DTFs, equalized between 50 Hz and 18 kHz for hi-fi auralizations<br />
* [http://sofacoustics.org/data/database/ari%20(altb) ARI (ALTB)]: HRTFs from the ARI database. Measurements for some of the listeners from the ARI database, repeated and evaluated a few years later, see [http://www.researchgate.net/publication/236111930_Sound_localization_in_individualized_and_non-individualized_crosstalk_cancellation_systems Majdak et al. (2013)]. <br />
* [http://sofacoustics.org/data/database/cipic CIPIC]: HRTFs from the [http://interface.cipic.ucdavis.edu/data/ CIPIC database]. 45 listeners, partially antropometric data available.<br />
* [http://sofacoustics.org/data/database/riec RIEC]: Far-field HRTFs from the [http://www.riec.tohoku.ac.jp/pub/hrtf/index.html RIEC] database of over 100 human listeners. ''(Credit: Kajni Watanabe, Japan)''<br />
<br />
<br />
HRTFs of artificial heads:<br />
* [http://sofacoustics.org/data/database/mit MIT-KEMAR]: HRTFs from [http://sound.media.mit.edu/resources/KEMAR/ MIT of the KEMAR] dummy head. Reference HRTFs used in many publications.<br />
* [http://sofacoustics.org/data/database/ari%20(artificial) ARI (ARTIFICIAL)] HRTFs of mannequins (dummy heads) measured at ARI using the same setup as for human listeners:<br />
** NH169: HRTFs, DTFs, and raw data of a printed head of the corresponding human listener<br />
** NH172: HRTFs, DTFs, raw and reference data of the dummy head Neumann KU 100. Also part of Club Fritz, see below.<br />
* [http://sofacoustics.org/data/database/fhk FHK]: HRTFs from the [http://www.audiogroup.web.fh-koeln.de/ku100hrir.html Fachhochschule Köln] of the dummy-head Neumann K100. Gapless data, high spatial resolution provided. ''(Credit: Benjamin Bernschütz, Germany).''<br />
* [http://sofacoustics.org/data/database/scut SCUT]: Near-field HRTFs from SCUT database of the KEMAR (Radius: 0.2 to 1.0 m). ''(Credit: Bosun Xie, China)''<br />
* [http://sofacoustics.org/data/database/tu-berlin TU-Berlin]: HRTFs from [https://dev.qu.tu-berlin.de/projects/measurements/repository/show/2010-11-kemar-anechoic/mat TU-Berlin of the KEMAR] dummy-head. HRTFs for several distances (>0.5 m). ''(Credit: Hagen Wierstorf, Germany)''<br />
* [http://sofacoustics.org/data/database/clubfritz Club Fritz]: HRTFs of Neumann KU 100 measured within the project Club Fritz where several institutions measured the same artificial head ''(Credit: Brian Katz, France)''. Coming soon...<br />
<br />
<br />
Special HRTFs:<br />
* [http://sofacoustics.org/data/database/ari%20(bte) ARI (BTE)]: Behind-the-ear HRTFs and DTFs from the [http://www.kfs.oeaw.ac.at/hrtf ARI database].<br />
<br />
<br />
DRIRs: <br />
* [http://sofacoustics.org/data/database/oldenburg Oldenburg] DRIRs from [http://medi.uni-oldenburg.de/hrir/html/download.html Oldenburg]. Recordings in an office under several conditions ''(Credit: Stephan Ewert, Germany)''.<br />
<br />
<br />
== Headphone impulse responses (HpIRs) ==<br />
'''New: HpIRs of humans: '''<br />
* [http://sofacoustics.org/data/headphones/ari ARI]: HpIRs from the ARI database. Single headphone, five measurements (with repositioned headphone in-between) for over 100 listeners.<br />
* [http://sofacoustics.org/data/headphones/btdei BT-DEI]: HpIRs from the [http://padva.dei.unipd.it/?page_id=345 BT-DEI] database provided. 16 Listeners, 3 headphones ''(Credit: Michele Geronazzo, Italy)''.<br />
<br />
<br />
== Special purpose: ==<br />
<br />
* [http://sofacoustics.org/data/amt amt]: HRTFs for the various models from the [http://amtoolbox.sourceforge.net AMToolbox]<br />
* [http://sofacoustics.org/data/sofa_api_mo sofa_api_mo] HRTFs as examples for the [http://sourceforge.net/projects/sofacoustics/ SOFA API for Matlab/Octave]<br />
<br />
<br />
== Other repositories ==<br />
<br />
This is a list of other repositories providing HRTFs, BRIRs, and DRIRs available as SOFA files.<br />
<br />
* '''ARI free-field HRTF database'''. HRTFs available for in-the-ear (ITE) and behind-the-ear (BTE) HRTFs. Link: http://www.kfs.oeaw.ac.at/hrtf<br />
* '''Example files''' created by the Matlab/Octave API. Link: http://tinyurl.com/sofaHRTFs<br />
* '''RIEC database''': Database from the Advanced Acoustic Information Systems Laboratory, Research Institute of Electrical Communication, Tohoku University, Japan. Link: http://www.riec.tohoku.ac.jp/pub/hrtf/index.html</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1866SimpleFreeFieldSOS2014-10-15T09:27:55Z<p>Tcarpent: /* Description */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data type ==<br />
<br />
=== SOS ===<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.SOS||<nowiki>[1 1]</nowiki>||m||mRn||double||Coefficients of the SOS filters (see below for details)<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.SOS<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||MR||double||Monaural delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
The 'Parents' attribute may be used to refer to the SimpleFreeFieldHRIR file (if any) that was the origin of the parametric model.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1865SimpleFreeFieldSOS2014-10-15T09:23:14Z<p>Tcarpent: /* Data types */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data type ==<br />
<br />
=== SOS ===<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.SOS||<nowiki>[1 1]</nowiki>||m||mRn||double||Coefficients of the SOS filters (see below for details)<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.SOS<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||MR||double||Monaural delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1864SimpleFreeFieldSOS2014-10-15T09:22:12Z<p>Tcarpent: /* Data types */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data types ==<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.SOS||<nowiki>[1 1]</nowiki>||m||mRn||double||Coefficients of the SOS filters (see below for details)<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.SOS<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||MR||double||Monaural delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1863SimpleFreeFieldSOS2014-10-15T09:20:37Z<p>Tcarpent: /* Data types */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data types ==<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.IR||<nowiki>[1 1]</nowiki>||m||mRn||double||Impulse responses<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.IR<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||MR||double||Monaural delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1862SimpleFreeFieldSOS2014-10-15T09:19:25Z<p>Tcarpent: /* Data types */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data types ==<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.IR||<nowiki>[1 1]</nowiki>||m||mRn||double||Impulse responses<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.IR<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||IR, MR||double||Monaural delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1861SimpleFreeFieldSOS2014-10-15T09:19:00Z<p>Tcarpent: /* Data types */</p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data types ==<br />
<br />
For storing second-order sections filters. <br />
<br />
Note: Delay is mandatory (set to 0 if not used). <br />
<br />
{| border="1"<br />
!Name<br />
!Default<br />
!Flags<br />
!Dimensions<br />
!Type<br />
!Comment<br />
|-<br />
|Data.IR||<nowiki>[1 1]</nowiki>||m||mRn||double||Impulse responses<br />
|-<br />
|Data.SamplingRate||<nowiki>48000</nowiki>||m||I||double||Sampling rate of the samples in Data.IR<br />
|-<br />
|Data.SamplingRate:Units||<nowiki>hertz</nowiki>||m||||attribute||Unit of the sampling rate<br />
|-<br />
|Data.Delay||<nowiki>[0 0]</nowiki>||m||IR, MR||double||Delay in samples<br />
|}<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SimpleFreeFieldSOS&diff=1860SimpleFreeFieldSOS2014-10-15T09:16:57Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
== Data types == <br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1856Talk:SOFA specifications2014-10-13T19:45:15Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization process (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=SOFA_conventions&diff=1855SOFA conventions2014-10-13T15:26:32Z<p>Tcarpent: </p>
<hr />
<div>SOFA conventions specify the description of data and metadata for a particular kind of measured data.<br />
<br />
This is important in order to meet the different requirements coming from different application fields, specified. For example, description of HRTFs, BRIRs, and DRIRs requires different metadata. Also, some applications may prefer to see the data stored in a different way. These conventions, once approved by the peer group are defined in SOFA conventions.<br />
<br />
==Stable SOFA Conventions==<br />
Stable SOFA conventions are those for which SOFA files are publicly available and can be read/modified by a software package. The following stable SOFA conventions are available now:<br />
* [[GeneralFIR]]: General convention with FIR as DataType (no restrictions but DataType)<br />
* [[GeneralTF]]: General convention with TF as DataType (no restrictions but DataType)<br />
* [[SimpleFreeFieldHRIR]]: Free-field HRTFs stored as impulse responses, measured with an omnidirectional source for a single listener.<br />
* [[SimpleHeadphoneIR]]: Conventions to store headphone IRs recorded for each emitter and each ear, single listener and no directionality of emitter/receiver considered.<br />
<br />
<br />
==Proposed SOFA Conventions==<br />
Proposed SOFA conventions are currently being discussed. We consider following rules for new SOFA conventions:<br />
* Data must exist (do not foresee the future)<br />
* Data can not be described by existing SOFA conventions<br />
* Relevant information about the data available<br />
<br />
<br />
The following SOFA conventions are being discussed. Measured data exist but their description must be fixed in order to create publicly available SOFA files and corresponding software interfaces.<br />
<br />
<br />
* [[SimpleFreeFieldTF]]: as SimpleFreeFieldHRIR, but uses TF as DataType covering special needs coming from HRTF simulations<br />
* [[SingleRoomDRIR]]: Room impulse responses measured with an arbitrary number of receivers (such as a microphone array) and an omnidirectional source in a single room.<br />
<br />
* ([[SimpleBRIR]]): Binaural room impulse responses measured with an omnidirectional source in a single reverberant space. Somebody wanted to have this, but the work stopped at the moment.<br />
<br />
== Unsorted topics ==<br />
Here we list the suggestions and feedback from the peer group:<br />
* Include anthropometric data<br />
* Crosstalk cancellation filters<br />
* Include calibration data from the measurement<br />
* Include room pictures<br />
<br />
Please use the "Discussion" function to discuss these topics.<br />
<br />
== Feedback and Contribution ==<br />
<br />
If you would like to contribute or propose new SOFA conventions: <br />
* Send an e-mail to the [mailto:sofacoustics-devel@lists.sourceforge.net mailing list]. You don't have to be a member of the mailing list to send a message to the moderator. <br />
* Go to one of the SOFA pages of your interest and use the "Discussion" for your contribution.<br />
We appreciate your feedback!</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1854Talk:SOFA specifications2014-10-13T14:47:11Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1853Talk:SOFA specifications2014-10-13T14:46:14Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p).<br />
<br />
The SimpleFreeFieldSOS does not impose any restriction on the filter coefficients contained in Data.SOS. Especially it is the user responsibility to check whether the filters are stable or not.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1852Talk:SOFA specifications2014-10-13T14:42:09Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p)<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1851Talk:SOFA specifications2014-10-13T14:41:49Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the cascaded filters are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
For each filter H(z), the filter taps are arranged as follows:<br />
<br />
[ B1(z) A1(z) B2(z) A2(z) .... Bp(z) Ap(z) ] == [ b10 b11 b12 a10 a11 a12 b20 b21 b22 a20 a21 a22 .... bp0 bp1 bp2 ap0 ap1 ap2 ]<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
(N = 6 * p)<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1850Talk:SOFA specifications2014-10-13T14:39:12Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the filters is the cascade are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
Each second order section is represented by 6 filter taps:<br />
<br />
Bi(z) = bi0 + bi1 z^(-1) + bi2 z^(-2)<br />
Ai(z) = ai0 + ai1 z^(-1) + ai2 z^(-2)<br />
<br />
(even though the denominator is usually normalized such that ai0 = 1).<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1849Talk:SOFA specifications2014-10-13T14:37:28Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the filters is the cascade are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
The SOS is represented as follows:<br />
<br />
H(z) = B1(z) / A1(z) . B2(z) / A2(z) . ... . Bp(z) / Ap(z)<br />
<br />
where p is the number of second order sections.<br />
<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).<br />
<br />
<br />
We encourage to use the 'History' attribute to store informations relative to the modelization (algorithm(s) used to derive the IIR model, method used to estimate the monaural delays, etc.)</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1848Talk:SOFA specifications2014-10-13T14:34:16Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
For the sake of simplicity, we consider that all the filters is the cascade are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
SimpleFreeFieldSOS convention : everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
DataType is 'SOS' (standing for 'second order sections').<br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (expressed in samples).</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1847Talk:SOFA specifications2014-10-13T14:33:13Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
Everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
For the sake of simplicity, we consider that all the filters is the cascade are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
<br />
<br />
DataType is 'SOS' <br />
<br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
<br />
Data.Delay is [M R] and contains the monaural delay (in the units of N i.e. in samples).</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1846Talk:SOFA specifications2014-10-13T14:32:57Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (e.g. Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.<br />
<br />
==Proposal for SimpleFreeFieldSOS ==<br />
Everything similar to SimpleFreeFieldHRIR except the following:<br />
<br />
For the sake of simplicity, we consider that all the filters is the cascade are second order filters (i.e. no first order filter). (anyway a first order filter can be represented with a second order filter).<br />
<br />
<br />
<br />
DataType is 'SOS' <br />
Data.SOS of size [ M R N ] contains the filter coefficients. N being the total number of coefficients, it is always a multiple of 6.<br />
Data.Delay is [M R] and contains the monaural delay (in the units of N i.e. in samples).</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1845Talk:SOFA specifications2014-10-13T14:26:27Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
<br />
Several perceptual studies (Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.</div>Tcarpenthttps://www.sofaconventions.org/mediawiki/index.php?title=Talk:SOFA_specifications&diff=1844Talk:SOFA specifications2014-10-13T14:25:39Z<p>Tcarpent: </p>
<hr />
<div>SimpleFreeFieldSOS : Simple Free Field Second Order Sections<br />
<br />
==Description==<br />
<br />
This convention is essentially the same as SimpleFreeFieldHRIR except that it is adapted to a parametric model of HRIR.<br />
This parametric model consists of replacing HRIR (as FIR) with a monaural delay and second order sections filters.<br />
<br />
==Background==<br />
Being a causal and stable filter, an HRIR can be decomposed into a minimum phase part component and an all-pass component.<br />
The phase of each HRIR is thus decomposed into the minimum phase and the phase of the all-pass component i.e. the excess phase.<br />
The minimum phase is related to the magnitude spectrum through the Hilbert transform.<br />
The excess phase of HRIR is usually linear (up to approx 8 - 10 kHz). <br />
A simplified model of HRTF can thus be built where the all-pass component is replaced by a pure delay. This pure delay is referred to as monaural delay.<br />
Each HRTF is thus fully described by its magnitude spectrum and the monaural delay.<br />
Several perceptual studies (Wightman 1992) have shown the validity of such simplified model (given the fact that the phase information of the higher frequencies -- that is neglected by the model -- is not used by the auditory system to estimate the directions of arrival).<br />
Finally the minimum phase part of HRTF can be modeled as an IIR filter. Many modeling techniques have been proposed. Anyway the resulting IIR digital filter needs to be represented as a cascade of first order or second order sections for numerical stability reasons.</div>Tcarpent