(Issue #246): Check of the Deep Neural Network package stored in the AMT configuration.
(Issue #256): Start-up time improvement by factor three. A complete re-write was required with code parts separation by toolbox installation and toolbox starts. A special starter for the LTFAT (overloading ltfatstart.m) added, needing only a few 100 ms even on Octave. The starter is versioned, with the current support for LTFAT 2.4.0 (October 2018) and LTFAT 2.5.0 (May 2022). For other LTFAT versions, the LTFAT's default starter is used.
(Issue #250): Dependency of amt_subdir removed.
Octave compatibility improved: io package is obligatory now, handling of packages simplified.
Octave compatibility: nan package can be optionally loaded, nanmean and nanstd can be used in Octave now.
(Issue #246): Support for Deep Neural Network package configuration.
(Issue #251): Textread replaced by fread because obsolete.
(Issue #234): All ° symbols replaced by deg to be parsed in figures correctly.
Figures better layouted, now using a table which combines the code, figure, and code output.
Distance between the options and their description increased.
Bug fix for loading files with a blank in the file name.
(Issue #272): Time out extended to 10 seconds and the ghosting .html files are removed after the download.
amt_extern: Support of empty outputs added.
changes (Issue #258): "Description" removed from the documentation.
amt_subdir (Issue #250): Removed because of two licenses required and dependencies have been removed.
amt_info (Issue #264): Dependence of amt_subdir removed and expanded to work on other files than models only.
amt_cache: Bug fix, disp replaced by amt_disp, doc improved.
amt_mex: Error on cleaning removed (clean.bat added in /environments).
mat2doc (Issue #273): Linking to ghosting figures removed.
verhulst2015: Adapted to use amt_extern.
eurich2022: Doc fixes, arg_eurich adapted to be AMT compliant.
auditoryfilterbank_singlefc.cc (Issue #259): Now compiles under Octave on Linux.
llado2022: llado2022_weightsanalysis removed because not a model stage, many fixes (code and docs).
decheveigne2023 (Issue #262): New model implementation and new argument function arg_decheveigne2023.
(Issue #277): Removed dependency on Image Processing Toolbox.
(Issue #139): Automatic pre-panning added to remove the parameter N in the model, see N_prepanning in the default parameters.
exp_engel2022 (Issue #245): Code updated to match the publication year. Changes include auxdata, cache, and the documentation.
exp_engel2021 (Issue #249): Moved to legacy.
exp_eurich2022: Doc fixes. Code improvements. Status improved to verified.
exp_llado2022: Check of Deep Neural Network Toolbox added, cached data if not available, auxdata in two versions now: with and without NNET Toolbox required.
(Issue #263): Fig. 5 is plotted in four separate panels now.
(Issue #247): Last row of Fig. 5 seems to be completly wrong, other rows not quite correct. exp_laback2023 refactored for better debugging, but the issue moved to later version.
demo_llado2022 (Issue #246): Checks of Deep Neural Network package improved; use cache if package not available; adapted for Octave compatibility.
plot_llado2022: Adapted for Octave compatibility.
demo_mckenzie2021 (Issue #239): Documentation improved, auxdata adapted to match the publication year .
plot_tabuchi2013 (Issue #224): Integrated in exp_tabuchi2013 because it was used only once and only there. plot_tabuchi2013 stays available in legacy for a while.
NEW: mat2doc.py directly integrated in the AMT (see /mat2doc/readme.txt)
NEW: upload.py script in /mat2doc/php to automatically adapt the documentation PHP files
amt_load: functions loading HRTFs use amt_load now. SOFAload and /hrtf are obsolete (Issue #231)
amt_load: support for CSV files, which are read by tableread
amt_extern: AMT can be run in a directory containing blanks now (Issue #233)
amt_version: AMT version is stored at a single place only and used by mat2doc and amt_configuration (Issue #219)
amt_configuration: display in silence mode fixed (Issue #226)
NEW: eurich2022: Binaural detection model based on interaural coherence
NEW: laback2023: Contextual lateralization based on interaural level differences
NEW: smalt2014: Medial olivocochlear reflex in auditory nerve responses
king2019: updated documentation
NEW: demo_dau1997 (Issue #189)
exp_osses2022: use of amt_disp instead of sprintf (Issue #180)
exp_osses2022: local functions called have prefix local (Issue #183)
exp_barumerli2021: moved to legacy (Issue #232)
exp_bauremrli2022: moved to legacy (Issue #232)
exp_majdak2013: added, but no figure reproduced on the website yet (Curve Fitting Toolbox required).
FIX: king2019 avoids NaN channels after scaling if there are empty channels
NEW: various model stages previewing the functionality of decheveigne2023 model: decheveigne2023_spikeach, decheveigne2023_spikecch, decheveigne2023_spikeisih, decheveigne2023_spikejitter, decheveigne2023_spikepoisson, decheveigne2023_spikepsth, decheveigne2023_spikerefractory, decheveigne2023_spiketopulse, decheveigne2023_spiketrain, decheveigne2023_spikevs
NEW: sig_laback2023: generates stimuli used in exp_laback2023
NEW: sig_kolarik2010: Tone masked by a diotic inner-band and two antiphasic flanking-band noises
NEW: sig_marquardt2009: Tone masked by a constant-ITD inner-band and two antiphasic flanking-band noises
NEW: sig_vanderheijden1999: Tone masked by a constant-ITD inner-band noise
king2019 avoids NaN channels after scaling if there are empty channels (Issue #235)
data_pausch2022: clean up (not completed yet)
license and copyright updated
SOFA 2.1 support
amt_start: do not require certificates when downloading toolboxes
EXTENSION: amt_load: supports SOFA format
lyon2011: ihc stage fixed
UPDATE: data_majdak2013: additional data for plotting Figure 6 from Majdak et al. (2013)
baumgartner2014_pmv2ppp: output optimized when 'print' used
exp_relanoiborra2019 : removed dependency from Matlab's Communication Systems toolbox
demo_llado2022 : removed dependency from Matlab's Communication Systems toolbox
FIX: amt_start does not change current directory anymore
FIX: amt_start('install') error caught when no compiler present
FIX: amt_stop fully restores settings prior to amt_start
FIX: amt_emuexp typos and char parsing
itdestimator: more robust 'levelthr' method: set ITD to NaN if no threshold found.
FIX: exp_baumgartner2021 Fig2, Fig3, Tab3
exp_baumgartner2021 restructured (renaming of local functions)
exp_breebaart2001 runs now more robustly under Linux
FIX: exp_lavandier2022 Fig5
NEW: barumerli2022: (renamed from deprecated barumerli2021) minor updates (new thirdparty function randvmf, removed dependency from Computer Vision Toolbox)
mckenzie2022: (renamed from deprecated mckenzie2021) experiment and auxdata updated
relanoiborra2019: default minspl changed from -34 to -33.9794 dB (dbspl(2e-7,,100)) to yield minlvl_lin=2e-7 in the adaptation loops (to better match publication)
ziegelwanger2014: model option 5 added: estimate TOAs by using the maximum on LP-filtered IRs.
baumgartner2017 (and related files) removed because replaced by the actually published baumgartner2021
Version 1.1.0 (December 2021)
removed all appearances of global SPL offset modifications (ltfatsetdefaults).
amt_configuration: added downloadpath to enhance search in previous versions functionality
amt_start: added version-dependent removal of ssh certificate
amt_load: restructured download display
FIX: adaptloop bug fix with signal scaling.
NEW: minspl for adaptloop, which is minlvl in dB (minlvl still can be used for backwards compatibility).
adaptloop: new parameter minspl, which is minlvl in dB. minlvl as linear amplitude can still be used.
ihcenvelope: flag minlvl changed to be ihc_minlvl to be clearly associated with ihcenvelope. No backwards compatibility provided.
ihcenvelope: flag ihctype renamed to ihc_type.
NEW: f2erbrate: ERB rate for a frequency according to Moore and Glasberg (1983)
NEW: erbrate2f: frequency from the ERB rate according to Moore and Glasberg (1983)
NEW: bmld: binaural masking level difference according to Culling et al. (2005)
demo_adaptloop: minor improvements
demo_mclachlan2021: minor changes
demo_bruce2018: better parameters showing better the results
exp_baumgartner2015binweight: Changes of the SPL offset removed (output looks like Fig. 5 from Baumgartner and Majdak, 2015)
exp_osses2022: adapted to the revised manuscript version.
exp_bruce2018: adapted to the changes in bruce2018.
exp_li2020: adjusted sequence of Figures so they correspond to publication
exp_baumgartner2021: minor adjustments
Models and Modelstages:
carney2015: new flag 'ic_hwr' (and 'no_ic_hwr') to optionally disable the halve-way rectification of the outputs
bruce2018: new flag 'specificSRautoTiming' to provide spontanous rates but calculate timing information for the auditory nerve fibers (used by exp_osses2022)
zilany2014: improvements in displaying
- calculation of AN, CN, and IC can be disabled if unused
- detailed output provided only when requested (faster computation but causes backwards compatibility issues)
- major documentation improvement
- fixed erbspace and do_xcorr
- new option to calculate auditory filtering via auditoryfilterbank_singlefc