X-Git-Url: https://git.aubio.org/?p=aubio.git;a=blobdiff_plain;f=ChangeLog;h=e03dec66d4b5009f564f85114c12c05550449769;hp=7023719db969cf1705b28aa14ef8d559aa06cbce;hb=a4df8aaa1b7e40ce1e165a82ff41864b1c6209ab;hpb=b0edd2d5ca66ee50cbd81d6028a47de5b4777e9c diff --git a/ChangeLog b/ChangeLog index 7023719d..e03dec66 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,835 @@ +2018-11-21 Paul Brossier + + [ Overview ] + + * VERSION: bump to 0.4.8 + * notes: new option release_drop (gh-203) + * spectral: new parameters added to filterbank and mfcc (gh-206) + * python: start documenting module (gh-73, debian #480018), improve build for + win-amd64 (gh-154, gh-199, gh-208) + * fixes: prevent crash when using fft sizes unsupported by vDSP (gh-207), + prevent saturation when down-mixing a multi-channel source (avcodec/ffmpeg) + + [ Fixes ] + + * avcodec: prevent saturation when down-mixing a multi-channel source, emit + a warning if compiling against avutil < 53 (gh-137), wrap long lines + * examples/: avoid hiding global and unreachable code + * fft: limit to r*2*n sizes, with r in [1, 3, 5, 15] (vDSP only) (gh-207) + * fft: fix reconstruction for odd sizes (fftw only) + * pvoc: add missing implementations for aubio_pvoc_get_hop/win + * mathutils: increase ln(2) precision of in freqtomidi/miditofreq + * wavetable: stop sets playing to 0, add dummy implementation for _load + + [ New features ] + + * src/musicutils.h: new aubio_meltohz, aubio_hztomel, with _htk versions + * src/spectral/filterbank.h: new set_mel_coeffs, set_mel_coeffs_htk, + set_power, and set_norm methods, improved set_triangle_bands + * src/spectral/mfcc.h: new set_scale, set_power, set_norm, set_mel_coeffs, + set_mel_coeffs_htk, set_mel_coeffs_slaney + * src/mathutils.h: new fvec_mul + * src/notes: new option release_drop to prevent missing note-offs (gh-203) + + [ Python module ] + + * fix: rounding to nearest integer in midi2note and freq2note + * general: supports code generation of setters with none or multiple + parameters + * documentation: add docstrings do fvec, cvec, source, sink, pvoc, frequency + conversion and level detection routines (gh-73, debian #480018) + * slicing: improve and document slice_source_at_stamps + * module: new note2freq function, recover error log when raising exceptions + on failed set_ methods, prevent cyclic import, coding style improvements + * demos: improve coding style, fix bpm_extract arguments + * MANIFEST.in: exclude *.pyc, improve patterns + + [ Documentation ] + + * doc/: use sphinx autodoc to load docstrings from aubio module, reorganize + python module documentation, add a note about double precision, use https + when possible + * src/spectral/: update Auditory Toolbox url, update copyright year + + [ Tools ] + + * aubionotes: add --release-drop option + * aubio: add --release-drop and --silence options to `aubio notes`, + workaround for -V to really show version (py2) + * aubiocut: add option --create-first to always create first slice + + [ Tests ] + + * tests/, python/tests: add tests for new methods, check source channel + down-mix, improve coverage + + [ Build system ] + + * Makefile: disable docs when measuring coverage, add branch coverage + option, add coverage_zero_counters target, improve html report + * waf: update to 2.0.12, improve wscript style, prevent shipping some + generated files + * python: always show compiler warnings when pre-processing headers, + workaround to fix code generation for win-amd64 (gh-154, gh-199, gh-208). + * continuous integration: add azure pipelines, update and improve + configurations for appveyor, circleci, and travis. + +2018-09-22 Paul Brossier + + [ Overview ] + + * VERSION: bump to 0.4.7 + * src/spectral/dct.h: add dct type II object with optimised versions + * src/io/, src/notes/, src/pitch: prevent crashes on corrupted files + * examples/: fix jack midi output, improve messages when jack disabled + * python/: add dct support, minor bug fixes tests and demos + * wscript: improve support for BLAS/ATLAS + + [ Library fixes ] + + * src/pitch/pitchyinfft.c: fix out of bound read when samplerate > 50kHz + thanks to @fCorleone (closes #189, CVE-2018-14523, debian #904906) + * src/notes/notes.c: bail out if pitch creation failed (see #188) + * src/io/source_wavread.c: + - also exit if samplerate is negative (closes #188, CVE-2018-14522, + debian #904907) + - add some input validation (closes #148 and #158, CVE-2017-17054, + debian #883355) + * src/io/source_avcodec.c: + - give up if resampling context failed opening (see #137, closes #187, + CVE-2018-14521, debian #904908) + - give up reading file if number of channel changes during stream (closes + #137, CVE-2017-17554, debian #884237) + - make sure libavutil > 52 before checking avFrame->channels (see #137) + - fix build with ffmpeg 4.0, thanks to @jcowgill (closes #168, #173) + - avoid deprecated call for ffmpeg >= 4.0 + * src/onset/onset.c: add dummy default parameters for wphase (closes #150) + + [ Tools ] + + * examples/parse_args.h: hide jack options if not available, improve error + message (closes #182) + * examples/utils.h: process_block returns void + * examples/utils.c: fix examples failing to send more than one JACK midi + event per frame, thanks to @cyclopsian (closes #201) + + [ New features ] + + * src/spectral/dct.h: add dct type II object with implementation factory + * src/spectral/dct_plain.c: add plain dct implementation + * src/spectral/dct_ooura.c: add ooura implementation + * src/spectral/dct_fftw.c: add fftw implementation + * src/spectral/dct_ipp.c: add ipp version + * src/spectral/dct_accelerate.c: add vdsp/accelerate dct + * tests/src/spectral/test-dct.c: check reconstruction works + * src/spectral/mfcc.c: use new dct to compute mfcc + + [ Library internals ] + + * src/aubio_priv.h: avoid hard-coded undefs, split BLAS and ATLAS support, + add vdsp scalar add and multiply + + [ Build system ] + + * wscript: + - add options to disable examples and tests + - detect includes for openblas/libblas/atlas + * scripts/get_waf.sh: bump to 2.0.11, verify signature if gpg available + * python/lib/gen_external.py: pass '-x c' to emcc only + + [ Python ] + + * python/lib/gen_code.py: add support for rdo methods + * python/tests/test_dct.py: add tests for new dct + * python/demos/demo_pitch_sinusoid.py: use // to yield an integer, fixing + demo on py3, thanks to @ancorcruz (closes #176) + * python/ext/py-musicutils.*: add shift(fvec) and ishift(fvec) + * python/tests/test_fvec_shift.py: add tests for shift() and ishift() + * python/lib/aubio/cmd.py: fix typo in comment + + [ Documentation ] + + * README.md, doc/statuslinks.rst: use latest for commits-since + * examples/parse_args.h: add yinfast to pitch algorithms + * doc/requirements.rst: add some blas documentation + * doc/requirements.rst: split media/optimisation libraries + * doc/develop.rst: fixed spelling error, thanks to Jon Williams (closes #161) + * doc/aubio{pitch,notes}.txt: add yinfast to list of pitch methods + + [ Continuous integration ] + + * .travis.yml: remove xcode8.2 builds, group osx, add alias pip=pip2 + * .appveyor.yml: upgrade pip first, always use python -m pip + +2017-10-02 Paul Brossier + + [ Overview ] + + * VERSION: bump to 0.4.6 + * src/spectral/fft.c, src/*.c: add support for Intel IPP (many thanks to + Eduard Mueller) + * wscript: add support for emscripten (thanks to Martin Hermant) + * src/pitch/pitchyinfast.h: new fast method to compute YIN algorithm + * src/pitch/pitchyin*.c: improve confidence measure, making sure its value + corresponds to the selected period (thanks to Eduard Mueller) + * python/lib/aubio/cmd.py: add `quiet`, `cut`, and `help` subcommands + + [ Library ] + + * src/aubio_priv.h: add missing aubio_vDSP_vclr (Eduard Mueller) + * src/io/source_avcodec.c: improve error message, prevent un-opened bracket, + no declaration after statements for older compilers, avoid unused variable + * src/mathutils.c: prevent segfault with Accelerate.framework (closes #58, + closes #102) + * src/spectral/phasevoc.h: add aubio_pvoc_set_window to change the windowing + function + * src/mathutils.c: add window type `ones` (no windowing) + + [ Python ] + + * python/demos/demo_tapthebeat.py: add a real-time example to play beats + using pyaudio + * python/lib/gen_external.py: improve parsing and syntax, use results in + emscripten build (Martin Hermant) + * python/lib/aubio/cmd.py: add option `-u` to `aubio pitch`, improve error + messages, add `quiet` subcommand (closes #124), improve syntax, add some + documentation, add `cut` and `help` subcommand, add silence and time format + options + * python/lib/aubio/cut.py: upgrade to argparse, set samplerate as needed + * python/demos/demo_yin_compare.py: add comparison of yin implementations + * python/demos/demo_wav2midi.py: add an example to create a midi from a + sound file using mido (closes: #134) + * python/demos/demo_bpm_extract.py: use argparse, use beats_to_bpm function + * python/ext/py-cvec.c: fix support for pypy by changing setters to return a + negative value on error (closes #17) + + [ Documentation ] + + * src/tempo/beattracking.h: fix typo (thanks to Hannes Fritz) + * doc/requirements.rst: fix broken link (thanks to @ssj71, closes #99) + * doc/aubiomfcc.txt: fix typo in 'coefficients' + + [ Tests ] + + * python/tests/tests_aubio_{cmd,cut}.py: add basic tests + * python/tests/test_filterbank*.py: ignore UserWarnings, clean-up, + improve get_coeff tests + + [ Build system ] + + * wscript: add support for emscripten, see scripts/build_emscripten + * scripts/get_waf.sh: update waf to 2.0.1, build waf from source tarball + * scripts/build_emscripten: update to build aubio.js + * Makefile: add coverage and coverage_report targets, run tests once + + [ Continuous integration ] + + * .travis.yml: add coverage report on osx + * .appveyor.yml: use msvc 14.0 (VS 2015) and scripts/get_waf.sh + * .coveragerc: add minimal python coverage configuration + +2017-04-10 Paul Brossier + + [Overview] + + * VERSION: bump to 0.4.5 + * src/io/source_avcodec.c: add support for libswresample + * aubio: new python command line tool to extract information + * src/onset/onset.c: add spectral whitening and compression, improve default + parameters + * this_version.py: use centralized script to get current version, adding git + sha when building from git repo (thanks to MartinHN) + + [Interface] + + * src/spectral/awhithening.h: add adaptive whitening + * src/{cvec,mathutils,musicutils}.h: add cvec_logmag, fvec_logmag, and fvec_push + * src/onset/onset.h: add aubio_onset_set_default_parameters to load optimal + parameters of each novelty function, _{set,get}_compression and + _{set,get}_awhitening to turn on/off compression and adaptive whitening + * src/spectral/specdesc.h: add weighted phase + + [Library] + + * src/onset/onset.c: improve default onset parameters (thanks to @superbock + for access to his evaluation database), see commit dccfad2 for more details + * src/pitch/pitch.c: avoid segfault when using invalid parameters + * src/temporal/biquad.c: fix biquad parameters initialization (thanks to + @jurlhardt) + + [Tools] + + * examples/aubio{onset,track}.c: add options --miditap-note and + --miditap-velo to set which midi note is triggered at onset/beat (thanks to + @tseaver) + * examples/aubioonset.c: show actual parameters in verbose mode + * examples/utils.c: improve memory usage to emit midi notes + + [Python] + + * python/ext/py-source.c: add with (PEP 343) and iter (PEP 234) interface + * python/ext/py-sink.c: add with interface (PEP 343) + * python/lib/aubio/cmd.py: new `aubio` command line tool + * python/lib/aubio/cut.py: moved from python/scripts/aubiocut + + [Documentation] + + * doc/*.rst: reorganize and improve sphinx manual + * doc/*.txt: update manpages, add simple manpage for aubio command line + * doc/full.cfg: derive from doc/web.cfg + * README.md: simplify and add contribute information + + [Build system] + + * wscript: prefer libswresample over libavsamplerate when available, use + current version in manpages, doxygen, and sphinx, update to newest waf + * setup.py: use entry_points console_scripts to generate scripts, use + centralized version from this_version.py, clean up + * python/lib/moresetuptools.py: detect if libswresample is available + +2017-01-08 Paul Brossier + + [ Overview ] + + * VERSION: bump to 0.4.4 + * src/utils/log.h: new function to redirect log, error, and warnings + * python/: AUBIO_ERR raises python exception, AUBIO_WRN to emit py warning + * doc/: add some documentation, fix errors in manpages + * wscript: new rules to build 'manpages', 'doxygen', and 'sphinx', new + --build-type= option (thanks to Eduard Mueller) + * src/notes/notes.h: add minioi and silence methods + * examples/: add --minioi (minimum inter-onset interval) option + * src/pitch/pitchyin.c: improve msvc compiler optimisations (thanks to + Eduard Mueller) + * python/, src/: improve error messages, fix minor memory leaks + * src/io/source_avcodec.c: improve compatibility with latest ffmpeg and with + older libav/ffmpeg versions + * python/demos/: new demos to capture microphone in real time + + [ Interface] + + * src/aubio.h: include utils/log.h + * src/utils/log.h: add new aubio_log_set_function to redirect log messages + * src/notes/notes.h: add aubio_notes_{get,set}_minioi_ms, add + _{get,set}_silence methods + + [ Library ] + + * src/aubio_priv.h: add AUBIO_INF to print to stdout with header, use new + logging function, add ATAN alias, add stdarg.h, move #include "config.h" + * src/{fmat,fvec}.c: avoid integer division + * src/pitch/pitchyin.c: [msvc] help compiler to optimize aubio_pitchyin_do + by giving it addresses for all arrays which are referenced in inner loops, + thanks to Eduard Mueller. + * src/pitch/pitch.c: declare internal functions as static, fail on wrong + method, warn on wrong unit, improve error messages, fix error string + * src/spectral/specdesc.c: return NULL if wrong mode asked, remove trailing + spaces + * src/onset/onset.c: return null and clean-up if new_aubio_specdesc failed, + fix error message + * src/notes/notes.c: use midi note to store pitch candidate, round to + nearest note, add a variable to define precision, fix out-of-bound write, + fix unset silence_threshold, fix error message + * src/spectral/ooura_fft8g.c: add cast to avoid conversion warnings, prefix + public function with aubio_ooura_ to avoid with other apps using ooura (e.g. + puredata), make internal functions static, + * src/spectral/fft.c: add message about fftw3 being able to do non-power of + two sizes, make calls to fftw_destroy_plan thread-safe, use prefixed + aubio_ooura_rdft + * src/spectral/phasevoc.c: fix error string + * src/temporal/resampler.c: throw an error when using libsamplerate with doubles + * src/io/ioutils.h: add functions to check samplerate and channels, use in sink_*.c + * src/io/source.c: add error message when aubio was compiled with no source, + only show error message from last child source_ + * src/io/source_avcodec.c: call avformat_free_context after + avformat_close_input, keep a reference to packet to remove it when closing + file, avoid deprecation warnings with ffmpeg 3.2, add backward compatibility + for libavcodec55, fix for old libavcodec54, use AV_SAMPLE_FMT_DBL when + compiling with HAVE_AUBIO_DOUBLE, fix missing samples in eof block, avoid + function calls before declarations, improve error messages, replace with new + context before closing old one, make sure s->path is set to null + * src/io/{source_wavread,sink_wavwrite}.c: declare internal functions as static + * src/io/source_wavread.c: fix bytes_read for JUNK headers, improve error + messages, initialize buffer, skip chunks until data is found, or abort, skip + junk chunk + * src/io/source_sndfile.c: add support for multi-channel resampling, set + handle to null after sucessful close, add missing floor in ratio comparison, + improve formatting + * src/io/sink.c: only show error message from last child sink_ + * src/io/sink_apple_audio.c: avoid crash on empty file name + * src/io/sink_sndfile.c: improve error message + * src/io/sink_{sndfile,wavwrite}.c: use AUBIO_MAX_CHANNELS, fix error message + + [ Documentation ] + + * README.md: update copyright dates, use https + * src/aubio.h: add some links to examples, use https + * src/pitch/pitch.h: add aubio_pitch_get_tolerance, add basic description of + unit modes + * src/notes/notes.h: add doxygen header + * src/spectral/fft.h: strip example path + * doc/*.rst: improve sphinx documentation + * doc/android.rst: add reference to it scripts/build_android + * doc/debian_packages.rst: added page on debian packages + * doc/python_module.rst: add demo_source_simple.py, add note on pip, add + `print(aubio.version)` + * doc/cli.rst: include command line manpages + * doc/cli_features.rst: add matrix of command line features + * doc/requirements.rst: add a note about --notests (closes #77), document + --msvc options, improve description of options + * doc/download.rst: added page on download + * doc/installing.rst: update + * doc/xcode_frameworks.rst: added page on xcode frameworks + * doc/**: use https://aubio.org + * doc/conf.py: use pyramid theme, update copyright, remove hardcoded path + * doc/web.cfg: exclude ioutils from doc + * doc/aubionotes.txt: document -M option (see #18), + * doc/aubioonset.txt: add documentation for -M, --minioi, improve threshold + description (thanks to Peter Parker), fix typo (onset, not pitch) + * doc/aubio*.txt: document -T/--timeformat option + + [ Build ] + + * Makefile: add a brief intro, avoid offline operations, add html and dist + targets, add rules for documentation, simplify listing, avoid offline + operations, bump waf to 1.9.6, check for waf before clean, chmod go-w + waflib, improve clean, use pip to install, factorise pip options, generate + more test sounds, improve test_python and test_pure_python, pass build_ext + in test_pure_python{,_wheel}, quieten uninstall_python if already + uninstalled, improve test targets, use bdist_wheel in test_pure_python, + build_ext only for --enable-double, verbose waf rules, add cleanwaf + * wscript: added debug/release build type configurations release (default) + enables optimizations, debug symbols are enabled in both configurations, + thanks to Eduard Mueller. + * wscript: add options to disable source_wavread/sink_wavwrite, add check + for stdarg.h, new rules 'manpages', 'sphinx', and 'doxygen' to build + documentation, add version to sphinx and manpages, disable libsamplerate + if double precision enabled (libsamplerate only supports float), fix typos, + remove trailing spaces, improve tarball creation (./waf dist), remove + full.cfg from tarball, prepend to CFLAGS to honor user cflags + * wscript, src/wscript_build: improve install locations using DATAROOTDIR, + MANDIR, INCLUDEDIR + * wscript: default to no atlas for now + * src/wscript_build: always build static library + * scripts/build_android: add an example script to build aubio on android, + + [ Tools ] + + * examples/aubionotes.c: use new notes, set minioi, send last note off when + needed, add warning for missing options + * examples/aubioonset.c: add minioi option, in seconds + * examples/: only send a last note off when using jack + * examples/: return 1 if object creation failed + * examples/: use PROG_HAS_OUTPUT, add PROG_HAS_SILENCE + + [ Tests ] + + * tests/src/spectral/test-fft.c: fix default size + * tests/src/spectral/test-phasevoc.c: fix typos + * tests/src/utils/test-log.c: add AUBIO_INF, add example for + aubio_log_set_function, improve messages + + [ Python ] + + * python/ext/aubiomodule.c: add aubio._aubio.__version__ and import it as + aubio.version, use custom logging function for errors and warnings, remove + duplicated add_generated_objects, use <> for non local aubio + * python/ext/py-cvec.c: use NPY_INTP_FMT + * python/ext/py-fft.c: use error string set in src/spectral/fft.c + * python/ext/py-phasevoc.c: use error string set in src/spectral/phasevoc.c + * python/ext/py-sink.c: always set samplerate and channels in init + * python/ext/py-source.c: use error string set in src/io/source.c + * python/lib/aubio/midiconv.py: add unicode double sharp and double flat, + improve unicode handling, skip UnicodeEncodeError on python 2.x + + [ Python build ] + + * MANIFEST.in: add src/**.c, exclude full.cfg, include waflib, remove + python/ext/config.h + * setup.py: define AUBIO_VERSION use sorted glob.glob to improve + reproducibility, remove extra quotes, remove status from version string, + update description, use custom build_ext instead of 'generate' command, + define HAVE_AUBIO_DOUBLE to 1 if needed + * python/lib/gen_code.py: add support for multiple _do outputs, fix number + of output, improve del_ function, safer DECREF, fix indentation, emit RuntimeError + * python/lib/gen_external.py: clean-up, enable tss, remove duplicate, + sort generated files + * python/lib/moresetuptools.py: add HAVE_STDARG_H, also check for + HAVE_AUBIO_DOUBLE, cleaner clean, look first for system library, then for + local build, then local sources, mo nore fake config.h here, use + samplerate in single precision only + * python/README.md: add a note about nose2 for python tests (closes #74) + * scripts/setenv_local.sh: python3 compat + + [ Python demos ] + + * python/demos/demo_alsa.py: add example using alsaaudio (closes #72) + * python/demos/demo_mfcc.py: add options to plot first and second + derivatives, and set samplerate/win_s/hop_s, thanks to @jhoelzl (closes #68) + * python/demos/demo_notes.py: add simple notes demos + * python/demos/demo_pyaudio.py: added simple demo for pyaudio, see #6, + closes #78, thanks to @jhoelzl and @notalentgeek, add some comments, avoid + overwriting aubio.pitch + * python/demos/demo_source_simple.py: fix indentation, make executable + * python/demos/demo_timestretch{,_online}.py: fix usage string, remove + unused import, use // to yield an integer (closes #71) + * python/demos/demo_timestretch_online.py: use 512, fix block counter + * python/demos/demo_tss.py: improve default parameters, exit before plotting + + [ Python tests ] + + * python/tests/: use local import, add __init__.py + * python/tests/test_cvec.py: simplify + * python/tests/test_fft.py: skip test fft(zeros).phas == 0 if needed, expected powerpc + * python/tests/test_fvec.py: reduce alpha norm precision to 10.-4 + * python/tests/test_{midi2note,note2midi}.py: use nose2.params, add unicode tests + * python/tests/test_notes.py: add basic tests + * python/tests/test_notes.py: test results are correct for 44100Hz_44100f_sine441.wav + * python/tests/test_sink.py: add more tests, quiet warnings + * python/tests/test_source.py: break long line, check the tail of the file + is non-zero on non silent test files, filter user warnings to avoid spamming + the console, only check if last frames are non silent on brownnoise (weak), + remove fragile brownnoise test, check duration on short files, use nose2 + params to process one sound file per test + * python/tests/test_specdesc.py: RuntimeError is now raised on wrong mode + * python/tests/utils.py: by default, use 5 seconds brownoise + + [ Only in git ] + + * .travis.yml: add debian dpkg-buildflags config, switch from precise to + trusty, sudo required, add ffmpeg on osx, add targets ios, iosimulator, + and osx noopt configs, bump to xcode8, add xcode8.2 config, mimick + build_apple_frameworks options, alway upgrade pip, add pip --version and + which pip after upgrading, remove --user, use expandwaf in install, remove + unused ARCH, shuffle order, remove duplicate, add missing opening quote, + use AUBIO_NOTESTS to build only lib on ios, add gitter webhook + * .appveyor.yml: fix path for windows+python 3.5, fix typo in path, make + nose2 tests verbose + +2016-08-16 Paul Brossier + + [ Interface ] + + * src/io/source.h, src/io/source_*.h: add _get_duration + * src/notes/notes.h: add basic notes object + * src/tempo/beattracking.{c,h}: add _get_period and _get_period_s + * src/mathutils.h: add fvec_ishift + * src/fvec.{c,h}: add fvec_weighted_copy + * src/tempo/tempo.{c,h}: add _get_period and _get_period_s, also add tatum, + a subdivision of the beat period, default to 4, implement get/set_delay + * src/**.{c,h}: use #ifdef HAVE_FOO, not #if _HAVE_FOO, add const qualifiers + to unmodified pointers (see #35) + + [ Library ] + + * src/{fmat,fvec,mathutils}.c: optimisations (using atlas or Accelerate when + available) for fvec_sum, fvec_mean, fvec_shift, aubio_level_lin, + fvec_set_all, fvec_zeros, fvec_weight, fvec_copy, fvec_weighted_copy, + fmat_vecmul + * src/aubio_priv.h: check for atlas cblas, use cblas_xswap, vDSP_dotpr, + protect SQR parameters, avoid redefining MIN/MAX, define PATH_MAX and PI + when needed, use _isnan on windows msvc 9 to avoid linking error, more + windows hacks + * src/mathutils.c: avoid for loop initial declarations [gcc], use + HAVE_ATLAS, use smpl_t for constants + * src/fmat.c: skip asserts + * src/spectral/{filterbank,mfcc}.c: use accelerated fmat_vecmul + * src/spectral/fft.c: fftw can be used odd length sizes, not Ooura, + factorise double / single flags, use memcpy + * src/spectral/phasevoc.c: fix arguments checks, return NULL when fft + creation failed , apply windowing for resynthesized grain, use ishift for + odd windows, fix scaling factors for correct reconstruction at 50 and 75% + overlap + * src/pitch/pitch.c: allow for silence == 0, improve error messages + * src/pitch/pitchmcomb.c: fix candidates sorting function, really comparing + current to next + * src/notes/notes.c: equivalent to previous examples/aubionotes.c results + * src/onset/onset.c: simplify selection of first onset, fix for "conversion + from 'smpl_t' to 'uint_t', possible loss of data" with msvc + * src/pitch/pitchmcomb.c: scan across all spectrum + * src/pitch/pitchyinfft.c: use fvec_weighted_copy + * src/{spectral/*.c,onset/*.c,tempo/*.c}: make sure win_size > 1 + * src/io/*.c: use custom defines for {source,sink}_apple_audio, take a copy + of const char* path + * src/io/source_avcodec.c: + - update to libav10, libavcodec 55.34.1 + - avoid deprecation warning, detect if we use ffmpeg or libav version + - check if the uri is a network stream using av_url_split, call + avformat_network_init() if needed + - check if we still need max_analyze_duration2 (closes #53, thanks to + @anthonylauzon) + * src/io/source_{avcodec,sndfile}.c: avoid modifying input param + * src/io/{sink,source,utils}_apple_audio.c: fix memory leak calling + CFRelease (closes #26, closes #27, and closes #28) + * src/io/sink_apple_audio.c: disable async mode for now, factorise code + * src/io/source_apple_audio.c: check out of bounds _seek, set s->path, quiet + * src/io/source_sndfile.c: fix crash, zero-pad output vector when + upsampling, use sf_read_double when compiling with AUBIO_DOUBLE, approximate + duration when resampling + * src/io/sink_sndfile.c: fix for double precision + * src/synth/sampler.c: fix typo, keeps a copy of uri + * src/tempo/tempo.c: do not write novelty function in output[1] + * src/temporal/resampler.c: make msvc happier adding a dummy variable + * src/temporal/filter.c: check parameters, fix filter_do_outplace to really + avoid modifying input + * src/utils/windll.c: add dll main entry point + + [ Python ] + + * General: + - new build system, new code generator + - Python 3 compatibility (#33), thanks to Nils Philippsen (@nphilipp) + - double precision compatibility + - simplify memory allocations, removed unneeded malloc/free calls + - fix memory leak (#49), check input sizes (#63) and output sizes (#60) + - improve indentation, clean up unused imports and variables + - fix comparison to None and to False + * setup.py: move from python/setup.py, add option to build libaubio inside + python-aubio (for instance with pip), add command 'generate' with option + '--enable-double', build with -Wdeclaration-after-statement -Werror + * python/ext/aubiomodule.c: fix PyMethodDef sentinel + * python/ext/aubioproxy.c: factorize input checks into + PyAubio_IsValidVector, fix windows c89 compilation, use npy_intp, not long + * python/ext: rewrite and simplify, safer and improved memory usage (#49), + improve error strings, verify actual object creation + * python/ext/py-source.c: added duration, check seek is not negative + * python/ext/py-musicutils.c: do not overwrite PyArg_ParseTuple messages + * python/lib/gen_code.py: new generator, switch to using custom PyObjects + instead of fvec, cvec, fmat, ready for double precision (defaults to single) + * python/lib/aubio__init__.py: use new aubio.float_type, make sure length is + not zero and float_type is imported + * python/lib/aubio/midiconv.py: fix instance checks, make sure midi2note + uses midi int (#33) + * python/lib/aubio/slicing.py: fix samplerate + * python/ext/aubio-types.h: add new_py_ functions to create PyObjects + instead of fvec_t, apply to generated and hard-coded objects + * python/lib/gen_external.py: improve compiler detection, fixes build on + windows (#55) + * python/lib/moresetuptools.py: helpers for windows and macos compilations + + [ Python demos ] + + * python/demos/demo_reading_speed.py: new reading speed tests, external + packages disabled by default + * python/demos/demo_timestretch.py: new timescale algorithm + * python/demos/demo_timestretch_online.py: new timescale algorithm (online + version) + * python/demos/demo_create_test_sounds.py: add script to create simple sound + files to test on using sox + * python/demos/demo_a_weighting.py: add simple demo for a_weighting + * python/demos/demo_filter.py: moved from _a_weighting + * python/demos/demo_mfcc.py: use n_coeffs + * python/demos/demo_bpm_extract.py: add exception type, avoid {} as default + argument value + * python/demos/demo_pysoundcard_*: update to pysoundcard 0.5.2 (closes #42) + * python/scripts/aubiocut: fix usage string output + + [ Python tests ] + + * python/tests/run_all_tests,*.py: switch to nose2, fix most prospect warnings + * python/tests/test_fvec.py: add test_pass_to_numpy, cope with accumulated + errors + * python/tests/test_cvec.py: simplify, add more tests + * python/tests/test_fft.py: more tests, fft.do to clash on wrong size + inputs, f.rdo input size, cvec is large enough, memory tests, avoid + VisibleDeprecationWarning + * python/tests/test_filterbank.py: check for wrong values, ValueError raised + * python/tests/test_filter.py: add tests + * python/tests/test_musicutils.py: simplify, check TypeError is raised + * python/tests/test_mfcc.py: more tests, check for wrong input size (see #63) + * python/tests/test_mathutils: fix test_miditobin test, can also raise + NotImplementedError (darwin) + * python/tests/test_note2midi.py: more tests, use unicode_literals + * python/tests/test_phasevoc.py: add a note about ocasional crash check + perfect reconstruction for overlap > 75% add 50% overlap test, fix duplicate + test name, add wrong sized input tests + * python/tests/test_sink.py: remove useless many_sinks_not_closed and cruft + * python/tests/test_source.py: simplify, quieten, skip tests if no test sounds + * python/tests/test_specdesc.py: check for wrong values, skip wrong name + test, use correct input size (see #63) + * python/tests/utils.py: try reopening the file is deleting it fails on windows + * python/VERSION: removed, use same VERSION file for libaubio and python-aubio + * MANIFEST.in: move from python/, update contents + * nose2.cfg: add minimal config, set multiprocess always-on=false (fixes + coverage, pass -N to speed up the tests) + + [ Tools ] + + * examples/*.c: add time format option + * examples/{aubioonset,aubiotrack}.c: also emit midi note, thanks to + @topas-rec (closes #62) + * examples/: use outmsg to print notes (fixes #8) + * examples/aubionotes.c: use new aubio_notes object + * examples/aubiotrack.c: enable -O and -t options, fix is_beat/is_silence + types + * examples/{parse_args,utils}.h: check in config.h if getopt.h was found, or + build without for msvc, more windows hacks + * examples/utils.c: change send_noteon to accept floating point midi note number + + [ Tests ] + + * tests/src/io/test-source_apple_audio.c: shorten long line + * tests/src/io/test-source_avcodec.c: use HAVE_LIBAV, closes #10 + * tests/src/temporal: avoid crash, clarify + * tests/src/tempo/test-tempo.c: tempo back to only one output + * tests/src/test-delnull.c: improve test, avoid segfaults + * tests/src/test-lvec.c: use AUBIO_LSMP_FMT + * tests/utils_tests.h: add VA_ARGS versions of variadic macros + * tests/utils_tests.h: also use custom srandom/random when compiling with + -std=c99 + * tests/utils_tests.h: make sure M_PI and RAND_MAX are defined + + [ Build ] + + * Makefile: set waf to 1.8.22 for now, new targets create_test_sounds, + build_python, test_python, clean_python, build_python3, clean_python3, + test_pure_python, test_pure_python_wheel, (use test_pure_* targets to build + without libaubio), use 'HAVE_DOUBLE=1 make' to build in double precision + * scripts/build_apple_frameworks: add script to build macosx and ios + frameworks (see #34, #43) + * scripts/build_emscripten: add script to build with emcc and co + * scripts/build_mingw: add script to cross-compile using mingw + * scripts/get_waf.sh: added simple script to fetch latest waf + * scripts/setenv_local.sh: set environment to run from built source tree + * scripts/setenv_local.sh: update to new python-aubio build location + * tests/wscript_build: do not install test programs + * tests/wscript_build, src/wscript_build: use 'use =', simplify + * src/wscript_build: enable shared lib on ios, static lib on windows + * wscript: + - update --enable-foo to fail if foo is not found + - add -mmacosx-version-min=10.4 on darwin + - add '-fembed-bitcode' on ios (closes #31), min to 6.1 + - make fat build, add option to not build with Accelerate framework + - add option to not build with CoreAudio/AudioToolbox + - add --disable-docs option + - add -lm detection + - pass HAVE_AUBIO_DOUBLE in compiler arguments + - first check for headers, make getopt.h and unistd.h optional + - check HAVE_AV* from ctx.env + - make msvc compiler quieter, add /MD and /D_CRT_SECURE_NO_WARNINGS + - check if we find atlas/cblas.h + - new build platform emscripten + - more cleanups and updates + + [ Only in git ] + + * .travis.yml: config for https://travis-ci.org/aubio/aubio + * .appveyor.yml: config for https://ci.appveyor.com/project/piem/aubio + * .landscape.yml: config for https://landscape.io/github/aubio/aubio + * conda recipes: see https://github.com/conda/conda-recipes#387 + * .gitignore: add python/tests/sounds and .egg-info + + [ General ] + + * src/: remove trailing spaces, improve doxygen strings, update copyrights, + fix typos + * src/onset/onset.h: fix description of get/set_delay functions + * src/spectral/mfcc.h: add link to reference implementation + * src/spectral/filterbank_mel.h: update reference url + * src/musicutils.h: update link to Bernardini's paper, improve doc + * doc/aubiomfcc.txt: add a note about the output + * doc/*.cfg: update to Doxygen 1.8.8 + * python/README.md: fix typo (thanks to Sam Alexander), document how to + build in a virtualenv (see #2) + * README.md: minor updates, link to python/README.md, switch to https + * VERSION: bump to 0.4.3 + +2015-08-01 Paul Brossier + + [ Interface ] + * src/onset/onset.h: add aubio_onset_get_silence(), fix description of + aubio_onset_get/set_delay functions + * src/tempo/tempo.h: add aubio_tempo_get_silence and aubio_tempo_get_threshold + + [ Library ] + * src/io/audio_unit.c: fix deprecation warning (closes #16) + * src/io/sink_apple_audio.c: avoid opening null path + * src/io/sink_sndfile.c: improve error messages, set nsamples after write + * src/io/sink_wavwrite.c: fail if samplerate is way too large, assume + windows is little endian to build with mingw32 + * src/io/source_apple_audio.c: check out of bounds _seek, set s->path, quiet + * src/io/source_avcodec.c: update to libav10, libavcodec 55.34.1 + * src/io/source_sndfile.c: cast to uint_t for now, clarify some variables + names, fix crash, zero-pad output vector when upsampling, improve seek errors and strings + * src/io/source_wavread.c: avoid orphan parenthesis , cast size_t to int to + avoid warning on win64 , improve seek errors processing , sync error message + * src/onset/onset.c: avoid doubled onset at start, check parameters + * src/pitch/pitch.c: add Hertz as valid unit string, check parameters + * src/pitch/pitchyin.c, src/pitch/pitchyinfft.c: avoid producing NaN on silence (closes #7) + * src/pitch/pitchyin.{c,h}: fix typo, developed + * src/spectral/phasevoc.c: fix argument checks, improve error messages + * src/synth/sampler.c: make sure blocksize > 0 + * src/tempo/beattracking.c: avoid segfault on silence (closes #24) + * src/tempo/tempo.c: add silence detection (closes #23), add old behavior in + examples/aubiotrack.c, check parameters + * src/*.c: fix some useless float conversion + + [ Python ] + * python/ext/aubiomodule.c: improve documentation + * python/ext/py-musicutils.c: add silence_detection (closes #21), + level_detection, db_spl, level_lin, complete window implementation + * python/ext/py-source.c: add channels, add seek (closes #9) + * python/demos/demo_a_weighting.py: add simple demo for a_weighting + * python/demos/demo_sink_multi.py: add multi channel version + * python/demos/, python/demos/: minor improvements, improve syntax (closes #19) + * python/setup.py: use setuptools, add numpy to install_requires + + [ Build ] + * waf: removed, now automatically downloaded with make, updated to 1.8.12 + * Makefile: add target to fetch waf and basic instructions + * .travis.yml: added template for travis continuous integration + * {examples,src,tests}/wscript_build: add external libs and uselib to targets + * src/wscript_build: do not hardocode install path + * wscript, src/io/*.c: use custom defines instead of __APPLE__ + * wscript: make fat build optional, remove iOS sdk versioning + + [ General ] + * examples/: use outmsg to print notes (fixes #8) + * examples/parse_args.h: improve short documentation (closes #12) + * doc/*.cfg: remove html timestamps to make build reproducible (see debian bug #779216) + * doc/aubiocut.txt: improve documentation + * doc/*.rst: add sphinx documentation basis with Makefile and conf.py + * README.md: minor updates + * python.old: removed old code + * tests/: minor bug fixing + +2014-01-31 Paul Brossier + + [ Interface ] + * src/io/{source_wavread,sink_wavwrite}.h: new source and sink objects to + read and write simple wav files when building with no external dependencies + * src/io/{source,sink}*.h: added close functions to explicitly close file + * src/io/sink*.h: added do_multi, preset_channels, preset_samplerate, + get_channels and get_samplerate functions + * src/onset/onset.h: added aubio_onset_get_threshold + * src/mathutils.h: add fvec_quadratic_peak_mag to find the magnitude of + interpolated peaks + + [ Library ] + * src/io/source_avcodec.c: implement _seek function + * src/io/source_sndfile.c: fixed _seek when signal is resampled + * src/spectral/ooura_fft8g.c: use float when double is not needed + * src/io/{source,sink}_apple_audio.c: improve error messages + * src/spectral/phasevoc.c: optimize swapbuffers by using memcpy + * src/lvec.c: add missing lvec_set_sample, improve test + * src/tempo/beattracking.c: improve confidence values by using interpolated + peak magnitude and avoiding nan + * src/tempo/tempo_davies.c: set default novelty function to specflux + + [ Python module ] + * python/ext/py-{sink,source}.c: add do_multi and close functions + * python/ext/aubio-types.h: set NPY_NO_DEPRECATED_API to version 1.7 + * python/lib/aubio/slicing.py: new function to slice a source into a list of + arbitrary regions + * python/scripts/aubiocut: add options --cut-until-nsamples and + --cut-until-nsclices, thanks to Mark Suppes for requesting, sponsoring, + and testing this feature + + [ General code ] + * src/aubio_priv.h: use ifdefs to avoid checking undefined defines + * src/**.c: add missing prototypes, avoid some declarations after statement + * waf, waflib: update to 1.7.15 + 2013-12-08 Paul Brossier * Overdue: After more than five years of development behind the curtain, @@ -10,7 +842,7 @@ * Memory management: allocation and freeing of memory has been optimized in many ways. Several memory leaks and out of bound access have been fixed. - * Optimization: the FFT, central to most algorithm, can now be computed + * Optimization: the FFT, central to most algorithms, can now be computed using different optimized algorithms, depending on what is available on your platform (FFTW, Ooura, or vDSP). Other simple optimization tricks are included. Most can be deactivated by configuring the build accordingly.