import dascore as dc
patch = dc.get_example_patch("sin_wav", sample_rate=100, duration=3, frequency=2)
spec = patch.dft("time", real=True, pad=False)
centroid = spec.spectral_centroid(fmin=1, fmax=3)
peak_frequency = spec.spectral_peak_frequency(fmin=1, fmax=3)
peak_amplitude = spec.spectral_peak_amplitude(fmin=1, fmax=3)Spectral Descriptors
Spectral descriptors summarize Fourier-domain patches. First transform a patch with dft or stft, then call descriptor methods directly on the resulting patch.
spectral_centroid is the power-weighted center frequency. spectral_peak_frequency returns the frequency bin with the largest spectral power, while spectral_peak_amplitude returns the largest linear spectral amplitude. Both peak descriptors select the global maximum within the requested frequency range, rather than finding multiple local peaks.
Other descriptors include median_frequency, spectral_entropy, spectral_kurtosis, and spectral_flatness. These methods accept DFT and STFT output and support frequency limits with fmin and fmax.
Descriptor outputs remove the frequency dimension and coordinates that depend on it, while preserving all remaining coordinates and their units.