- Removed SWIPE' algorithm
- It is not based on autocorrelation, I skipped it in all of the tests, and my implementation was basically copy-pasted from kylebgorman/swipe: just use their code instead!
- Fix autocorrelation (in YIN and MPM) for power-of-two sizes in FFTS (see anthonix/ffts#65) by using r2c/c2r transforms (addresses #72 reported by @jeychenne)
- Fix PYIN bugs to pass all test cases (addresses @jansommer's comments in #84 (comment))
- Added many more unit tests, all passing (228/228)
Contributors
jeychenne and jansommer