hsercanatli / tonicidentifier_makam Goto Github PK

pypeaks doesnt work

• I/O
• code comments
  etc.

Explain the details.

Optimize the peak detection algorithm parameters. For some examples It does not find all the peaks perfectly.

Which licence type should be selected?

The set should be restricted to certain input type/range for each propery, when applicable

Also fix relevant functions (e.g. plot_tonic)

Put some extracted pitch track examples(json) to the rep.

The commented octave correction was ad hoc and it was giving erroneous results because it was not making the tonic octave finding correctly. Also the current assumption (i.e. the biggest peak in the tonic pitch class will give the octave) is false for many makams, where tonic is visited only in the end.

We need a more general approach:

1. Compute the overall distribution of the recording
2. Find all the octave candidates from the peaks in the tonic pitch class
3. Get the summation of all the values in the distribution starting from the candidate tonic to the octave of the candidate for each tonic candidate.
4. the highest summation will give the correct tonic