alexanderlerch / pyaca Goto Github PK

also add instantaneous frequency: alexanderlerch/ACA-Code@432c136

Originally posted by @alexanderlerch in #17 (comment)

als add nmf: alexanderlerch/ACA-Code@9a6f61a

Originally posted by @alexanderlerch in #17 (comment)

sequential forward selection: alexanderlerch/ACA-Code@5de6bbe

Originally posted by @alexanderlerch in #17 (comment)

see matlab commit alexanderlerch/ACA-Code@0075bba

LOOCV: alexanderlerch/ACA-Code@5de6bbe

Originally posted by @alexanderlerch in #17 (comment)

also add pca: alexanderlerch/ACA-Code@9384b96

Originally posted by @alexanderlerch in #17 (comment)

replace scipy spectrogram function (who would come up with this absurd function signature?) with self-made block by block np fft

specifically kNN, LOO-CV, Feature Selection, see the matlab implementations here: alexanderlerch/ACA-Code@5de6bbe

see https://github.com/alexanderlerch/ACA-Code/blob/master/ComputeFingerprint.m

also add kmeans: alexanderlerch/ACA-Code@0f6edb1

Originally posted by @alexanderlerch in #17 (comment)

according to this Matlab commit alexanderlerch/ACA-Code@3193585#diff-af851cfa40a96d2785d0b587c423130aa50923d29c064271268a51fc83a2230c

Hi Professor Alexander,

There is a little oversight in the code that may be confusing when running the sample code, not a problem though.

In the function computeFeatureCl https://github.com/alexa nderlerch/pyACA/blob/cd1bcad2f0fee38037ef4e9b9929e03a95ae8491/pyACA/computeFeature.py#L111

def computeFeatureCl(cPath, cFeatureName, bPlotOutput = False):

    # read audio file
    [f_s, afAudioData] = ToolReadAudio(cPath)
    afAudioData = np.sin(2*np.pi * np.arange(f_s*1)*440./f_s)

The test code was not commented so 'afAudioData' is always overwritten after reading from a wav file.

Thanks for your work!

Sincerely,
Zhe

Check for consistent variable/function/file naming, order of function parameters, and order of return values.

Ensure that helper functions (e.g., ToolDownmix} are used where relevant to avoid code duplication

compare alexanderlerch/ACA-Code@aa1e5d3

incorporate the matlab code changes from here: https://github.com/alexanderlerch/ACA-Code/blob/master/ComputeBeatHisto.m

freq2bark, freq2mel, freq2midi, midi2freq

I'm trying to install pyACA in a venv but I can't succed to install pyACA.

The erros is :
pip install https://github.com/alexanderlerch/pyACA.git
Collecting https://github.com/alexanderlerch/pyACA.git
Downloading https://github.com/alexanderlerch/pyACA.git
/ 162 kB 2.4 MB/s
ERROR: Cannot unpack file /private/var/folders/23/_3qwnfqc8xj09j006s6mmcgr0000gp/T/pip-unpack-7hnq4j91/pyACA.git (downloaded from /private/var/folders/23/_3qwnfqc8xj09j006s6mmcgr0000gp/T/pip-req-build-768hmqvn, content-type: text/html; charset=utf-8); cannot detect archive format
ERROR: Cannot determine archive format of /private/var/folders/23/_3qwnfqc8xj09j006s6mmcgr0000gp/T/pip-req-build-768hmqvn

All the best
Yan

see alexanderlerch/ACA-Code@57f7ab7

Dear Professor Lerch,

I used one of your scripts,
computepitch, in order to list F0 values and time stamps,

Would it be possible that for values that should be zero, or in Praat tool 'undefined', extremely high values are generated. Do you have an idea, what the issue is ?

Thanks

Best regards

also add gmm: alexanderlerch/ACA-Code@0032bad

Originally posted by @alexanderlerch in #17 (comment)

incorporate the changes from here: https://github.com/alexanderlerch/ACA-Code/blob/master/NoveltyFlux.m

add knn: alexanderlerch/ACA-Code@5de6bbe

Originally posted by @alexanderlerch in #17 (comment)

... to keep pyACA in sync with the matlab repo:
modify FeatureTimeRms according to
alexanderlerch/ACA-Code@b1538a6#diff-f9ce41ae61b4fd3efe44ff1255954a4ef4c0bef3419ed872f5b1524f8a8eb1a8