librosax.feature.metrogram

metrogram(*, tg: ndarray, freqs: ndarray, factors: ndarray | None = None, aggregate: Callable | None = <function sum>, kind: str = 'linear', fill_value: float = 0) → ndarray

Summarize the presence of rhythmic ratios in a tempogram.

A tempogram with simultaneous energy at 90 BPM and 30 BPM has a strong 1/3 ratio, which makes meter estimation possible by looking for the ratio between the beat and downbeat frequencies. The default factors 1/3, 1/4, 1/5 and 1/7 correspond to 3/4, 4/4, 5/4 and 7/4 time.

This is a wrapper around librosa.feature.metrogram.

Parameters

tgnp.ndarray [shape=(…, n_freqs, n)]

Tempogram, as produced by tempogram or fourier_tempogram

freqsnp.ndarray [shape=(n_freqs,)]

Frequencies (in BPM) of the tempogram rows

factorsnp.ndarray or None

Rhythmic ratios to evaluate. Defaults to [1/3, 1/4, 1/5, 1/7].

aggregatecallable or None

Aggregation over the ratio axis. If None, the un-aggregated product is returned with the ratio axis at -3.

kindstr

Interpolation kind, as in scipy.interpolate.interp1d

fill_valuefloat

Value used outside the interpolation range

Returns

metrogramnp.ndarray

Ratio energies with shape (…, n_factors, n). If aggregate is None the un-aggregated product is returned instead, with the tempo axis retained: (…, n_factors, n_freqs, n).

See Also

tempogram fourier_tempogram librosa.feature.metrogram