Releases: nico-franco-gomez/dsptoolbox
Releases · nico-franco-gomez/dsptoolbox
v0.4.7
v0.4.7
Added
- new
dft
intransforms
for computing DFTs with any resolution lpc
intransforms
ExponentialAverageFilter
infilterbanks
- support for python 3.13
Misc
- improved precision of parallel filter by adding a third feed-forward
coefficient to least-squares approximation - replaced convolve with oaconvolve in multiple places for optimal handling
with different signal lengths - made framed signal methods available in
dsptoolbox.tools
- general doc corrections and additions
- added numba as new dependency for parallelizing some functions. It will be
installed and used automatically if the current python environment is 3.12 or
below. Support for numba and python 3.13 is not yet available.
Bugfix
- fixed problem with group delay designer
- fixed a problem with array dimensions in autoregressive coefficients estimation
v0.4.5
0.4.5
Added
FilterChain
infilterbanks
for use in real-time applicationsarma
infilterbanks
for obtaining arbitrary IIR filter approximations
to an impulse response
Misc
- renamed smoothe to smoothing across the library
- zeros, poles and gain are now saved in
Filter
. They are returned instead
of recomputing from the coefficients - general doc improvements
Bugfix
- corrected a bug where the time window of an impulse response did not match
after some time-domain operation was applied to it - fixed a problem with normalization in
audio_io
- fixed a problem with
Distortion
ineffects
v0.4.4
0.4.4
Added
- bark and erb approximations to warping factor
ParallelFilter
infilterbanks
KautzFilter
infilterbanks
- Realtime capabilities for filter
LatticeLadderFilter
,StateVariableFilter
,
IIR
,FIR
,KautzFilter
warp_filter
intransforms
resample_filter
instandard
Misc
- moved
kautz
andkautz_filters
functionality toKautzFilter
Bugfix
- use the peak for
combine_ir_with_dirac
instead of delay with minimum-phase
v0.4.3
0.4.3
Added
- added
laguerre
totransforms
- added
kautz
andkautz_filters
totransforms
Misc
- energy decay curve is not corrected with compensation energy or pruned from
noise when something during the estimation goes wrong (fallback strategy) - updated README
- moved
warp_ir
totransforms
and renamed towarp
- general documentation additions and fixes
- finding the end of an IR now also allows for defining a distance to noise
floor
Bugfix
- fixed a bug during the computation of the energy decay curve where
phase-inverted peaks were not taken into account for the start of the
impulse response
v0.4.2
0.4.2
Added
apply_gain
utility function instandard
- beta parameter for arbitrary noise generation
GroupDelayDesigner
infilterbanks
- nomalization of signals now accepts rms values
Misc
- frequency response interpolation with more interpolation modes
- refactored
PhaseLinearizer
Bugfix
- corrected a case where scaling of spectrum while plotting was wrong
v0.4.1
v0.4.0
0.4.0
Added
ImpulseResponse
as a subclass ofSignal
. It handles time windows, coherence
and plotting of those windows. Assertions for expectedImpulseResponse
instead
ofSignal
were added as well- new module
tools
for computations with primitive data types, added time
smoothing, interpolation of frequency response get_transfer_function
in Filter and FilterBank- analog-matched biquads in
filterbanks
gaussian_kernel
approximation infilterbanks
- gain parameter functionality for some biquads
- new biquad types (lowpass and highpass first order, inverter)
- new explicit constructors for signal and filter
- pearson correlation as part quality estimator for latency computation
- new scaling parameter in synchrosqueezing of
cwt
- new parameter in
window_frequency_dependent
Bugfix
- bugfix in
window_frequency_dependent
when querying a single frequency bin - corrected plotting of spl when calibrated signal is passed
Misc
- got rid of signal type attribute. Use now
ImpulseResponse
- general doc additions and fixes, type annotations
fractional_octave_smoothing
performance improved- renamed some files of code base for consistency
v0.3.9
0.3.9
Added
pinking_filter
infilterbanks
module
Bugfix
- fixed framed signal representation such that the last frames that need zero-padding
can be left out - biquad filter coefficients now use double precision by default
- minor fix in
window_frequency_dependent
Misc
- added zero-padding while computing minimum phase ir for better results
- compatibility with numpy v2.0 has been ensured
v0.3.8
Misc
- renamed paramater
remove_impulse_delay
toremove_ir_latency
- changed default values in
PhaseLinearizer
- general documentation improvements
Bugfix
find_ir_latency
now searches for the latency in comparison to the minimum
phase irharmonic_distortion_analysis
was fixed so that it can succesfully trim
the fundamental ir