Releases: ltfat/ltfat
Releases · ltfat/ltfat
2.6.0
Matlab
On Windows download the ltfat-[version]-win64.zip.
Download the ltfat-[version]-src.tgz when working on a different system than Windows.
To start the toolbox run
ltfatstart
GNU Octave
Download ltfat-[version]-src-of.tar.gz, and install it manually using
pkg install ltfat-[version]-of.tar.gz
from within the Octave prompt. Note that this only installs the toolbox. In order to make ltfat available run
pkg load ltfat
From the ChangeLog:
Version 2.6.0 (September 2023)
* NEW:
- gabreassignreal
- comp_gabreassignreal
- gabphasederivreal
- demo_waveletfilters
- lowdiscrepancy
- thirdparty/im2double (Octave function)
- ltfatruntests
- fusionframe
- arg_fusionframe
* BUGFIX:
- comp_fblowpassfilters
- audfilters
- gabphasederiv
- freqwavelet
- ltfat_mex_template_helper
- oct/filterbanks
- test_all_ltfat
- test_dft
- test_dgt_alg
- test_gabphasederiv
- test_gabphasederivinterface
- test_gabphasederivreal
- test_idft
- test_ptpfun
* EXTENSIONS: filterbanktight and filterbankrealtight (force painless option), frame (wavelet, auditory, warped, and gabor filter bank), waveletfilters (help example)
* ADJUSTMENTS: gabfilters and waveletfilters output format (to match that of the other filterbank generators)
* DOCUMENTATION: comp_dgt_real, comp_sepdgtreal, comp_gabreassign, comp_transferfunction,
demo_auditoryfilterbank, demo_filterbanks, filterbankconstphase, filterbankwin, framebounds
* DEPRECATED: comp_gdgt, test_dgt_fac, test_dgts, test_dwilts, test_framemulappr, test_gabmul, test_gdgt, test_hermbasis, test_ola, test_pherm, test_wfac
---------------------------
2.5.0
Matlab
On Windows download the ltfat-[version]-bin-win64.zip.
Download the ltfat-[version]-src.tgz when working on a different system than Windows.
To start the toolbox run
ltfatstart
From the ChangeLog:
Version 2.5.0 (May 2022)
* Improved precison of Daubechies wavelet filters for 'db1'-'db38'
Thanks to KIMURA Masaru.
* Added internal reset condition to 'multidgtrealmp'
Thanks to Zdenek Prusa.
* New filterbank-generating function 'waveletfilters'
Thanks to Nicki Holighaus.
* Extended filterbankdual and filterbankrealdual with an 'economize_filters' option
* Addition of 'comp_fblowpassfilters'
* Extended wavelet function generator (supported Morlet and several others)
freqwavelet, determine_freqatheight, arg_freqwavelet, wpghi_findalpha, helper_waveletgeneratorfunc
* New demo_dgt_parametrize
* rms accepts now 'dim' as an input
* frana accepts now row vectors as input signals
* expchirp optional fade-in and fade-out
* Bug fixes and minor changes:
nsdgt: Thanks to Shane F. Carr
normalize, filterbankrealbounds, cqtfilters, firkaiser, gabwin, comp_frana_tensor
* Deprecated: normalize - replaced by setnorm
2.4.0
Matlab
On Windows download the ltfat-[version]-bin-win64.zip or ltfat-[version]-bin-win64-R2018a.zip if your Matlab version is R2018a or higher.
Download the ltfat-[version]-src.tgz when working on a different system than Windows.
To start the toolbox run
ltfatstart
GNU Octave
From within the Octave prompt run
pkg install ltfat -forge
which will download and install ltfat from the official Octave forge site.
The ltfat-[version]-src-of.tar.gz is the identical version, which can be installed manually using
pkg install ltfat-[version]-src-of.tar.gz
Note that this only installs the toolbox. In order to make ltfat available run
pkg load ltfat
From the ChangeLog:
Version 2.4.0 12/10 2018
* Added universal filterbank phase reconstruction function:
filterbankconstphase
Thanks to Nicki Holighaus.
* Added time-frequency jigsaw puzzle tonal-transinet-residual separation:
tfjigsawsep, plottfjigsawsep, demo_tfjigsawsep
Thanks to Daniel Haider.
* Added coiflet filters up to K=17.
Thanks to KIMURA Masaru.
* Added wavelet function generator (supported Cauchy, Morse)
freqwavelet
* Function cqtfilters now accepts windows from freqwin.
* MEX files are now compatible with the new MATLAB 2018a API.
* Deprecated cqt, icqt, erblett ierblett. Replacements:
cqt: cqtfilters and filterbank
icqt: filterbankdual and ifilterbank or ifilterbankiter
erblett: audfilters with 'erb' and filterbank
ierblett: filterbankdual and ifilterbank or ifilterbankiter ChangeLog
2.3.1
From the Changelog
- multidgtrealmp: Introduced 2 additional MP variants
- Fixed Matlab 2018a MEX compilation issue
- Bug fixes:
multidgtrealmp: 'pedanticsearch' option now works correctly with 'freqinv'
ifilterbank: Fixed the multichannel issue
Removed symbolic link project.bib preventing successful installation in Octave 4.4 on Windows.
Re-enabled compilation of polyboolclipper in the Octave package.
2.3.0
Matlab
Windows binary packages win32 and win64 are intended for 32bit and 64bit Matlab installations respectively.
Download the ltfat-[version]-src.tgz when working on a different system than Windows.
GNU Octave
From within the Octave prompt run
pkg install ltfat -forge
which will download and install ltfat from the official Octave forge site.
The ltfat-[version]-src-of.tar.gz is the identical version, which can be installed manually using
pkg install ltfat-[version]-src-of.tar.gz
From the ChangeLog
* New function for computing matching pursuit in DGT: multidgtrealmp
* Matching pursuit and orthogonal matching pursuit for general frames
(for small problems): franamp
* ltfat_inferno is now the default colormap.
2.2.1
From the ChangeLog
* Added optional debiasing to franalasso.
* Some adjustments in audfilters.
* Abandoned nargchk and narginchk. They are deprecated in recent Matlab.
* Fixed char to number conversion warning in wavload occurring on Octave
2.2.0
Form the Changelog:
* New Gabor window generating functions: ptpfun, ptpfundual, pebfun, pebfundual
* New filterbank functions: filterbankscale, ifilterbankiter
* New window and filter generating functions: freqwin, freqfilter
* New functions for changing DGT phase convention: phaselockreal, phaseunlockreal
* Included new colormaps: ltfat_inferno, ltfat_magma, ltfat_plasma,
ltfat_viridis, ltfat_nicecolormap
* erbfilters and audfilters can now generate Gammatone and other filters
* Removed deprication warnings in test_all_ltfat
* Mulaclab no longer requires the Image processing toolbox to be installed (for the
basic functionality).
* Mulaclab now works on GNU Octave (thanks to Florent Jaillet).
2.1.2
From the ChangeLog
- Improved functions for DGT and DGTREAL phase reconstruction: constructphase, constructphasereal
- The backend library is now in separate repository: https://github.com/ltfat/libltfat and it is included as a subtree in subdirectory libltfat/ (replacing src/)
2.1.1
From the ChangeLog
- New function for computing higher order phase derivatives: gabphasederiv
- New function doing the adjustable reassignment: gabreassignadjust
- New function for the Gabor transform phase reconstruction: constructphase, constructphasereal
- New filterbank-generating function: audfilters
- New generic quadratic TF distribution function: quadtfdist
- New functions for reading and writing wav files
(since wavread and wavwrite were removed in Matlab 2015b): wavload, wavsave
(taken from VOICEBOX http://www.ee.ic.ac.uk/hp/staff/dmb/voicebox/voicebox.html) - Added 'nojava' flag to ltfatstart to optionally disable adding the
blockproc JAVA classes to the classpath. - Updated the lBFGS method in frsynabs according to the new paper.
- Fixed bug in gabphasegrad which prevented it from being used with any
window other than Gaussian. - Fixed bug in comp_ifilterbank_fftbl for non-painless filterbanks.
2.1.0
From the ChangeLog
- New filterbank-generating function: warpedfilters
- New filterbank spectrogram reassignment function: filterbankreassign
- ltfatarghelper is now also implemented as a MEX file.
- The block function can now handle synchronized simultaneous playback and
recording. E.g. {'rec','file.wav'} plays a wav file while recording.
This is different from the option 'playrec', which works the "other way"
i.e. it records and play the recorded. - Function framebounds(...,'iter') now uses pcg to represent the inverse of
the frame operator when calculating the lowest eigenvalue using eigs. - Added implementation of some quadratic time-frequency representations:
Rihaczek, Wigner-Ville - Fixed #112 (Sourceforge): Octave package now compiles on Windows (mxe version)
thanks to John Donoghue. - Related to #111 (Sourceforge): Wavelet filters were hand-centered around the origin.
- comp_filterbank, comp_ifilterbank and ltfatarghelper can be compiled as MEX files on Octave.