librosax.feature.cqt2010

cqt2010(y: ~jax.Array, *, sr: float = 22050, hop_length: int = 512, fmin: float | None = None, fmax: float | None = None, n_bins: int = 84, bins_per_octave: int = 12, tuning: float = 0.0, filter_scale: float = 1.0, norm: float | None = 1.0, sparsity: float = 0.01, window: str = 'hann', scale: bool = True, pad_mode: str = 'reflect', res_type: str | None = None, dtype: ~numpy.dtype = <class 'jax.numpy.complex64'>, output_format: str = 'magnitude', earlydownsample: bool = True) → Array[source]

Compute constant-Q transform using the 2010 algorithm with multi-resolution.

This implementation follows nnAudio’s CQT2010v2 algorithm which is more memory-efficient than CQT1992. It creates a small CQT kernel for the top octave and uses downsampling to compute lower octaves.

Parameters:

  • y – Audio time series

  • sr – Sampling rate

  • hop_length – Number of samples between successive CQT columns

  • fmin – Minimum frequency (default: C1 = 32.70 Hz)

  • fmax – Maximum frequency (default: inferred from n_bins)

  • n_bins – Number of frequency bins

  • bins_per_octave – Number of bins per octave

  • tuning – Tuning offset in fractions of a bin

  • filter_scale – Filter scale factor

  • norm – Normalization type for basis functions

  • sparsity – Sparsification factor (not implemented)

  • window – Window function

  • scale – If True, scale the output

  • pad_mode – Padding mode

  • res_type – Resampling type (not used)

  • dtype – Complex data type

  • output_format – Output format (‘magnitude’, ‘complex’, ‘phase’)

  • earlydownsample – If True, use early downsampling optimization

Returns:

CQT matrix