Skip to main content

predominant_melody

Function predominant_melody 

Source
pub fn predominant_melody(
    salience_frames: &[f32],
    n_frames: usize,
    n_bins: usize,
    f_min_hz: f32,
    bins_per_semitone: f32,
    out_f0: &mut [f32],
) -> Result<usize, AudioError>
Expand description

Estimate the predominant fundamental frequency of each frame.

salience_frames is a flat row-major buffer of shape n_frames × n_bins (salience_frames[f * n_bins + b]) on the logarithmic grid where bin b is f_min_hz * 2^(b / (12 * bins_per_semitone)). For each frame the strongest bin is found, parabolically refined, and converted to Hz; frames whose peak salience is below VOICING_RATIO of the whole excerpt’s peak salience abstain with out_f0 = 0.0.

Returns the number of voiced frames (non-zero out_f0).

§Errors