Harmonic-percussive-source-separation is a human-made audio signal processing method that decomposes a mixed recording into harmonic and percussive components using time-frequency median filtering and masking. Parameters: analysis window length and hop size for STFT, median filter lengths along frequency and time axes for harmonic and percussive masks, soft-masking threshold, and reconstruction method. Persistence mechanism: documented in signal processing literature, implemented in open-source libraries such as librosa, and used in music information retrieval workflows; the algorithm persists through code libraries and reproducible processing pipelines. [formal: separatio harmonica percussiva | substrate: behavior | horizon: as-long-as-us | explicit: yes | epoch: 0.02]
Accepted ontology entry
harmonic-percussive-source-separation
Harmonic-percussive-source-separation is a human-made audio signal processing method that decomposes a mixed recording into harmonic and percussive components using time-frequency median filtering and masking. Parameters: analysis window l…
Definition
Why it is in scope
A human-made audio signal processing technique built to persist as a documented method for decomposing mixed recordings into harmonic and percussive streams.
Names and aliases
- harmonic-percussive-source-separationen · CANONICAL
Relations from this entry
- cmsqg4mko005f3e32eyp15c4sDEPENDS_ON →
Harmonic-percussive-source-separation operates on STFT magnitude spectrograms to build harmonic and percussive masks. Remove STFT and the method has no time-frequency representation to filter and ceases to operate.
- cmsp6zgox04bkjlssq1q4peafDEPENDS_ON →
HPSS applies windowing before STFT to compute spectrograms. Without windowing, spectral leakage would corrupt the harmonic/percussive mask boundaries. Law 8b: removal-test passes.
- cmsrzn8aj00sdh7yuok1j2u7iINSTANCE_OF →
Harmonic-percussive-source-separation is a specific technique for separating audio mixtures into harmonic and percussive components. Source-separation is the general class of methods for decomposing mixtures into constituent sources. Specific→general, nearest kind.
- cmsq4obya07m0jlssh92wv3v0DEPENDS_ON →
HPSS decomposes audio into harmonic and percussive components using time-frequency median filtering on the spectrogram — a spectral representation derived from STFT. Remove spectral-representation and the median filtering of magnitude spectra has no input to operate on. Present-tense dependency per Law 8.
Relations to this entry
No accepted relations in this direction.
Record identity
- Created
- Aug 24, 2026, 9:11 PM UTC
- Content hash
- f73f2b75c88858e49f8c2db15b24447dfabccbc770902c7b0df6f274f7b92609