a library for audio and music analysis
GPL-3.0 License
Bot releases are visible (Hide)
Version 0.4.9 includes:
src/
: improve overall stability, fixing potential crashes and memory leaks on invalid arguments (closes gh-216); improve library messages and reporting of system errorstests/
: major clean-up (closes gh-219), check return codes, increase code coveragepython/tests/
: switch to pytest (closes gh-163), check emitted warningspython/
: add pages to manual with brief descriptions of classesSpecial thanks to Guoxiang Niu (@niugx) for reporting the following vulnerabilities, fixed with this release:
Also many thanks to @dvzrv, @romanbsd, @Bee-HN, @Lord-Kamina, and everyone who helped reporting the issues solved with this release.
Published by piem almost 6 years ago
Version 0.4.8 includes:
src/notes
: new option release_drop (gh-203)src/spectral
: new parameters added to filterbank and mfcc (gh-206)python/lib
: start documenting module (gh-73, debian #480018), improve build forsrc
fixes: prevent crash when using fft sizes unsupported by vDSP (gh-207),Thanks to @mj15003, @OneDirection9, @notalentgeek, @ulyssesp, @MaxHastings, @baozix, and everyone who reported and helped tracking down these issues.
Published by piem about 6 years ago
Version 0.4.7 includes:
src/io/
, src/notes/
, src/pitch
: prevent crashes on corrupted filessrc/spectral/dct.h
: add dct type II object with optimised versionsexamples/
: fix jack midi output, improve messages when jack disabledpython/
: add dct support, minor bug fixes tests and demoswscript
: improve support for BLAS/ATLASMany thanks to @jcowgill, @WIZARDISHUNGRY, @ancorcruz, @cyclopsian, and @fCorleone for their patches, bug reports, and comments, and to everyone who contributed to this release.
See full ChangeLog for the detail.
Published by piem about 7 years ago
Version 0.4.6 includes:
yinfast
, a new version of the YIN pitch detection algorithm, that uses spectral convolution to compute the same results as the original yin
, but with a cost O(N log(N))
, making it much faster than the plain implementation (O(N^2)
)
Intel IPP optimisations (thanks to Eduard Mueller), available for Linux, MacOS, Windows, and Android
improved support for emscripten (thanks to Martin Hermant), which compiles the aubio library as a javascript module and lets you run aubio's algorithms directly from within a web-page.
0.4.6 also comes with several bug fixes and improvements.
Many thanks to Eduard Mueller (@emuell), Martin Hermant (@MartinHN), Hannes Fritz (@hztirf), Stuart Axon (@stuaxo), Jörg (@7heW4yne), ssj71 (@ssj71), Andreas Borg (@borg), Rob (@mlrobsmt) and everyone else for their valuable contributions and input.
Published by piem over 7 years ago
A new version of aubio, 0.4.5, is available.
This version features:
aubio
python command line tool to extract information from sound filesNew options --miditap-note
and --miditap-velo
have been added to aubioonset
and aubiotrack
to adjust the note and velocity of the midi note emitted by onsets and beats.
0.4.5 also comes with a bunch of fixes, including improved documentation, build system fixes, and platform compatibility.
Many thanks to Martin Hermant (@MartinHN), Sebastian Böck (@superbock), Travis Seaver (@tseaver) and others for their help and contributions.
Published by piem over 7 years ago
This version features a complete rewrite of the python module, making it faster and more memory efficient than any previous version. It also adds compatibility for Python 3.x.
The build system has been improved so that python-aubio can now be installed with pip using a simple command:
$ pip install aubio
0.4.3 also comes with a number of improvements and several optimisations, so you should see both the library and the python module run significantly faster. Many computations have been vectorised, using either Atlas/cblas or the Accelerate framework, depending on the host platform.
Published by piem over 7 years ago
This version features a new logging module that allows redirecting errors, warnings, and other messages coming from libaubio. As usual, these messages are printed to stderr
or stdout
by default.
Another new feature is the --minioi
option added to aubioonset
, which lets you adjust the minimum Inter-Onset Interval (IOI) separating two consecutive events. This makes it easier to reduce the number of doubled detections.
New demos have been added to the python/demos
folder, including one using the pyaudio module to read samples from the microphone in real time.