Parameters

Parameter classes configure spectrogram computations and audio feature extraction.

STFT Parameters

class spectrograms.StftParams

Bases: object

STFT parameters for spectrogram computation.

centre

Whether to centre frames with padding.

hop_size

Hop size between frames.

n_fft

FFT size.

window

Window function.

Spectrogram Parameters

class spectrograms.SpectrogramParams

Bases: object

Spectrogram computation parameters.

classmethod music_default()

Create default parameters for music processing.

Uses n_fft=2048, hop_size=512, Hanning window, centre=true

Parameters

sample_ratefloat

Sample rate in Hz

Returns

SpectrogramParams

SpectrogramParams with standard music settings

sample_rate

Sample rate in Hz.

classmethod speech_default()

Create default parameters for speech processing.

Uses n_fft=512, hop_size=160, Hanning window, centre=true

Parameters

sample_ratefloat

Sample rate in Hz

Returns

SpectrogramParams

SpectrogramParams with standard speech settings

stft

STFT parameters.

Decibel Conversion

class spectrograms.LogParams

Bases: object

Decibel conversion parameters.

floor_db

Minimum power in decibels (values below this are clipped).

Frequency Scales

Mel Scale

class spectrograms.MelParams

Bases: object

Mel-scale filterbank parameters.

f_max

Maximum frequency in Hz.

f_min

Minimum frequency in Hz.

n_mels

Number of mel bands.

norm

Filterbank normalization strategy.

ERB Scale

class spectrograms.ErbParams

Bases: object

ERB-scale (Equivalent Rectangular Bandwidth) filterbank parameters.

f_max

Maximum frequency in Hz.

f_min

Minimum frequency in Hz.

n_filters

Number of ERB filters.

Constant-Q Transform

class spectrograms.CqtParams

Bases: object

Constant-Q Transform parameters.

num_bins

Total number of CQT bins.

Audio Features

Chromagram

class spectrograms.ChromaParams

Bases: object

Chromagram (pitch class profile) parameters.

f_max

Maximum frequency in Hz.

f_min

Minimum frequency in Hz.

classmethod music_standard()

Create standard chroma parameters for music analysis.

tuning

Tuning frequency in Hz (typically 440.0 for A4).

MFCC

class spectrograms.MfccParams

Bases: object

MFCC (Mel-Frequency Cepstral Coefficients) parameters.

n_mfcc

Number of MFCC coefficients.

classmethod speech_standard()

Standard MFCC parameters for speech recognition (13 coefficients).