Acoustic feature extraction (pitch, spectral flux, roll-off), audio representations (MFCC, Mel-spectrogram), and equalization DSP. More...

Files
file	pitch_harmonics.hpp
	Fundamental-frequency estimation and harmonic-content metrics.
file	audio_io.hpp
file	freq_scales.hpp
file	cepstral.hpp
file	melspec_delta.hpp
file	shape_descriptors.hpp
	Spectral-shape descriptors.

Classes
struct	nk::audio::AudioData
	Decoded audio waveform data and basic metadata. More...
struct	nk::audio::AudioInfo
	Audio file metadata container (info = audioinfo(filename)). More...

Functions
Value	nk::audio::pitch (const Value &x, double fs, double minF=50.0, double maxF=400.0, std::pmr::memory_resource *mr=nullptr)
	Pitch via the Normalised Cross-correlation Function (the default estimator).
Value	nk::audio::pitchCEP (const Value &x, double fs, double minF=50.0, double maxF=400.0, std::pmr::memory_resource *mr=nullptr)
	Pitch via the cepstrum method (Noll 1967).
Value	nk::audio::pitchPEF (const Value &x, double fs, double minF=50.0, double maxF=400.0, std::pmr::memory_resource *mr=nullptr)
	Pitch Estimation Filter (Gonzalez & Brookes 2011).
Value	nk::audio::pitchLHS (const Value &x, double fs, double minF=50.0, double maxF=400.0, std::pmr::memory_resource *mr=nullptr)
	Log Harmonic Sum / Subharmonic Summation (Hermes 1988).
Value	nk::audio::pitchSRH (const Value &x, double fs, double minF=50.0, double maxF=400.0, std::pmr::memory_resource *mr=nullptr)
	Summation of Residual Harmonics (Drugman & Alwan 2011).
Value	nk::audio::harmonicRatio (const Value &x, double fs, std::pmr::memory_resource *mr=nullptr)
	Harmonic ratio — strength of harmonic content per frame.
AudioData	nk::audio::audioreadFromBytes (const std::uint8_t data, std::size_t len, int64_t startSample=1, int64_t endSample=-1, bool nativeType=false, std::pmr::memory_resource mr=nullptr)
	Decodes audio from raw memory buffer, sniffing format automatically ([y, Fs] = audioread(bytes)).
AudioData	nk::audio::audioreadFromBytes (const std::string &bytes, int64_t startSample=1, int64_t endSample=-1, bool nativeType=false, std::pmr::memory_resource *mr=nullptr)
	Decodes audio from string memory buffer (audioread(bytes)).
AudioInfo	nk::audio::audioinfoFromBytes (const std::uint8_t *data, std::size_t len)
	Reads audio metadata from memory buffer without decoding audio samples (info = audioinfo(bytes)).
AudioInfo	nk::audio::audioinfoFromBytes (const std::string &bytes)
	Reads audio metadata from string memory buffer (info = audioinfo(bytes)).
std::vector< std::uint8_t >	nk::audio::audiowriteToBytes (const Value &y, double sampleRate, const std::string &format="wav", uint16_t bitsPerSample=16, const std::string &title="", const std::string &artist="", const std::string &comment="")
	Encodes audio matrix y into memory byte buffer (bytes = audiowrite(y, Fs, ...)).
AudioData	nk::audio::audioread (const std::string &path, int64_t startSample=1, int64_t endSample=-1, bool nativeType=false, std::pmr::memory_resource *mr=nullptr)
	Reads audio file from filesystem ([y, Fs] = audioread(filename)).
AudioInfo	nk::audio::audioinfo (const std::string &path)
	Returns audio file metadata (info = audioinfo(filename)).
void	nk::audio::audiowrite (const std::string &path, const Value &y, double sampleRate, uint16_t bitsPerSample=16, const std::string &title="", const std::string &artist="", const std::string &comment="")
	Writes audio data matrix y to file on disk (audiowrite(filename, y, Fs)).
Value	nk::audio::hz2mel (const Value &hz, std::pmr::memory_resource *mr=nullptr)
	Hertz → Mel (O'Shaughnessy 1987 default form).
Value	nk::audio::mel2hz (const Value &mel, std::pmr::memory_resource *mr=nullptr)
	Mel → Hertz (inverse of hz2mel).
Value	nk::audio::hz2bark (const Value &hz, std::pmr::memory_resource *mr=nullptr)
	Hertz → Bark (Traunmüller 1990 with low/high-frequency corrections).
Value	nk::audio::bark2hz (const Value &bark, std::pmr::memory_resource *mr=nullptr)
	Bark → Hertz (inverse of hz2bark).
Value	nk::audio::hz2erb (const Value &hz, std::pmr::memory_resource *mr=nullptr)
	Hertz → ERB (Glasberg & Moore 1990).
Value	nk::audio::erb2hz (const Value &erb, std::pmr::memory_resource *mr=nullptr)
	ERB → Hertz (inverse of hz2erb).
Value	nk::audio::phon2sone (const Value &phon, bool standardIs532_2=false, std::pmr::memory_resource *mr=nullptr)
	Phon → Sone — loudness-level to loudness conversion.
Value	nk::audio::sone2phon (const Value &sone, bool standardIs532_2=false, std::pmr::memory_resource *mr=nullptr)
	Sone → Phon — loudness to loudness-level conversion.
Value	nk::audio::cepstralCoefficients (const Value &S, int numCoeffs=13, std::pmr::memory_resource *mr=nullptr)
	Generic cepstral coefficient extractor.
std::tuple< Value, Value, Value >	nk::audio::mfcc (const Value &x, double fs, int numCoeffs=13, std::pmr::memory_resource *mr=nullptr)
	Mel-frequency cepstral coefficients (MFCC).
std::tuple< Value, Value, Value >	nk::audio::gtcc (const Value &x, double fs, int numCoeffs=13, std::pmr::memory_resource *mr=nullptr)
	Gammatone cepstral coefficients (GTCC).
std::tuple< Value, Value, Value >	nk::audio::melSpectrogram (const Value &x, double fs, int numBands=32, std::pmr::memory_resource *mr=nullptr)
	Mel-band power spectrogram.
Value	nk::audio::audioDelta (const Value &x, int windowLength=9, std::pmr::memory_resource *mr=nullptr)
	Locally-smoothed first derivative along the time axis.
Value	nk::audio::spectralCentroid (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral centroid — power-weighted mean frequency.
Value	nk::audio::spectralSpread (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral spread — power-weighted std around the centroid.
Value	nk::audio::spectralRolloffPoint (const Value &x, const Value &f, double percentile=0.95, std::pmr::memory_resource *mr=nullptr)
	Spectral rolloff frequency.
Value	nk::audio::spectralDecrease (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral decrease — slope-like measure relative to first bin.
Value	nk::audio::spectralSlope (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral slope — least-squares slope of (f, X).
Value	nk::audio::spectralFlux (const Value &x, const Value &f, double p=2.0, std::pmr::memory_resource *mr=nullptr)
	Spectral flux — p-norm of frame-to-frame spectrum differences.
Value	nk::audio::spectralCrest (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral crest — max(X) / mean(X) per frame.
Value	nk::audio::spectralEntropy (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral entropy — Shannon entropy of normalised spectrum.
Value	nk::audio::spectralFlatness (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral flatness — geomean(X) / mean(X).
Value	nk::audio::spectralKurtosis (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral kurtosis — X-weighted 4th central moment.
Value	nk::audio::spectralSkewness (const Value &x, const Value &f, std::pmr::memory_resource *mr=nullptr)
	Spectral skewness — X-weighted 3rd central moment.

Detailed Description

Acoustic feature extraction (pitch, spectral flux, roll-off), audio representations (MFCC, Mel-spectrogram), and equalization DSP.

Function Documentation

◆ audioDelta()

Value nk::audio::audioDelta	(	const Value &	x,
		int	windowLength = 9,
		std::pmr::memory_resource *	mr = nullptr )

Locally-smoothed first derivative along the time axis.

Implements the standard delta-coefficient filter: \( \Delta x[n] = \frac{\sum_{k=1}^{M} k \cdot (x[n+k] - x[n-k])}{2 \sum_{k=1}^{M} k^2} \) where M = floor(windowLength / 2).

Parameters

x	Feature matrix (rows = time, cols = channels).
windowLength	Filter window length (odd, ≥ 3). Default 9.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of delta coefficients.

See also: melSpectrogram, mfcc

◆ audioinfo()

AudioInfo nk::audio::audioinfo ( const std::string & path )

Returns audio file metadata (info = audioinfo(filename)).

Parameters

path	Path to audio file.

Returns: AudioInfo struct containing format, channels, sample rate, and duration.

See also: audioread, audiowrite

◆ audioinfoFromBytes() [1/2]

AudioInfo nk::audio::audioinfoFromBytes ( const std::string & bytes )

Reads audio metadata from string memory buffer (info = audioinfo(bytes)).

Parameters

bytes Audio file bytes as string.

Returns: AudioInfo metadata struct.

◆ audioinfoFromBytes() [2/2]

AudioInfo nk::audio::audioinfoFromBytes	(	const std::uint8_t *	data,
		std::size_t	len )

Reads audio metadata from memory buffer without decoding audio samples (info = audioinfo(bytes)).

Parameters

data	Pointer to raw audio bytes.
len	Length of buffer in bytes.

Returns: AudioInfo containing file format, sample rate, channels, and duration.

◆ audioread()

AudioData nk::audio::audioread	(	const std::string &	path,
		int64_t	startSample = 1,
		int64_t	endSample = -1,
		bool	nativeType = false,
		std::pmr::memory_resource *	mr = nullptr )

Reads audio file from filesystem ([y, Fs] = audioread(filename)).

Parameters

path	Path to audio file.
startSample	1-based start frame index (default: 1).
endSample	1-based end frame index (default: -1 for all).
nativeType	True to return integer types, false for normalized [-1.0, 1.0] double.
mr	Memory resource for output allocation.

Returns: AudioData containing decoded samples matrix and metadata.

See also: audiowrite, audioinfo

◆ audioreadFromBytes() [1/2]

AudioData nk::audio::audioreadFromBytes	(	const std::string &	bytes,
		int64_t	startSample = 1,
		int64_t	endSample = -1,
		bool	nativeType = false,
		std::pmr::memory_resource *	mr = nullptr )

Decodes audio from string memory buffer (audioread(bytes)).

Parameters

bytes	Byte string containing audio file contents.
startSample	1-based start sample index.
endSample	1-based end sample index (-1 for end of stream).
nativeType	True to return integer types, false for normalized [-1.0, 1.0] double.
mr	Memory resource for output allocation.

Returns: AudioData containing audio matrix and sampling rate.

◆ audioreadFromBytes() [2/2]

AudioData nk::audio::audioreadFromBytes	(	const std::uint8_t *	data,
		std::size_t	len,
		int64_t	startSample = 1,
		int64_t	endSample = -1,
		bool	nativeType = false,
		std::pmr::memory_resource *	mr = nullptr )

Decodes audio from raw memory buffer, sniffing format automatically ([y, Fs] = audioread(bytes)).

Parameters

data	Pointer to raw audio file bytes.
len	Length of buffer in bytes.
startSample	1-based start sample index.
endSample	1-based end sample index (-1 for end of stream).
nativeType	True to return integer types, false for normalized [-1.0, 1.0] double.
mr	Memory resource for output allocation.

Returns: AudioData containing audio matrix and sampling rate.

◆ audiowrite()

void nk::audio::audiowrite	(	const std::string &	path,
		const Value &	y,
		double	sampleRate,
		uint16_t	bitsPerSample = 16,
		const std::string &	title = "",
		const std::string &	artist = "",
		const std::string &	comment = "" )

Writes audio data matrix y to file on disk (audiowrite(filename, y, Fs)).

Parameters

path	Path to output audio file (e.g. .wav, .aiff, .au).
y	Audio matrix N x C.
sampleRate	Sampling rate in Hz.
bitsPerSample	Quantization bit depth (default 16).
title	Optional track title metadata.
artist	Optional artist metadata.
comment	Optional comment metadata.

See also: audioread, audioinfo

◆ audiowriteToBytes()

std::vector< std::uint8_t > nk::audio::audiowriteToBytes	(	const Value &	y,
		double	sampleRate,
		const std::string &	format = "wav",
		uint16_t	bitsPerSample = 16,
		const std::string &	title = "",
		const std::string &	artist = "",
		const std::string &	comment = "" )

Encodes audio matrix y into memory byte buffer (bytes = audiowrite(y, Fs, ...)).

Parameters

y	Audio sample matrix N x C.
sampleRate	Sampling rate in Hz.
format	Audio format string ("wav", "aiff", "au").
bitsPerSample	Bit depth (16, 24, 32).
title	Optional metadata title.
artist	Optional metadata artist.
comment	Optional metadata comment.

Returns: Vector of encoded file bytes.

◆ bark2hz()

Value nk::audio::bark2hz	(	const Value &	bark,
		std::pmr::memory_resource *	mr = nullptr )

Bark → Hertz (inverse of hz2bark).

Parameters

bark	Bark-scale values.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of Hz values.

See also: hz2bark

◆ cepstralCoefficients()

Value nk::audio::cepstralCoefficients	(	const Value &	S,
		int	numCoeffs = 13,
		std::pmr::memory_resource *	mr = nullptr )

Generic cepstral coefficient extractor.

Pipeline: log10 / cubic-root rectification → DCT-II → keep first numCoeffs coefficients. Input S is L × M (filterbank bands × frames). Output is M × numCoeffs (frames first).

Parameters

S	L × M real filterbank energy matrix.
numCoeffs	Number of cepstral coefficients to keep. Default 13.
mr	Memory resource (nullptr → process default).

Returns: M × numCoeffs DOUBLE matrix.

See also: mfcc, gtcc

◆ erb2hz()

Value nk::audio::erb2hz	(	const Value &	erb,
		std::pmr::memory_resource *	mr = nullptr )

ERB → Hertz (inverse of hz2erb).

Parameters

erb	ERB-scale values.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of Hz values.

See also: hz2erb

◆ gtcc()

std::tuple< Value, Value, Value > nk::audio::gtcc	(	const Value &	x,
		double	fs,
		int	numCoeffs = 13,
		std::pmr::memory_resource *	mr = nullptr )

Gammatone cepstral coefficients (GTCC).

Same STFT + cepstralCoefficients pipeline as mfcc but with an ERB-spaced Patterson-Holdsworth gammatone filterbank (Slaney 1993): cascaded 4-stage biquad frequency response, FrequencyRange = [50, fs/2], NumFilters = ceil(hz2erb(fs/2) - hz2erb(50)).

Parameters

x	Real 1-D audio signal.
fs	Sample rate in Hz.
numCoeffs	Number of cepstral coefficients. Default 13.
mr	Memory resource (nullptr → process default).

Returns: Tuple (coeffs, delta, deltaDelta).

See also: mfcc, hz2erb

◆ harmonicRatio()

Value nk::audio::harmonicRatio	(	const Value &	x,
		double	fs,
		std::pmr::memory_resource *	mr = nullptr )

Harmonic ratio — strength of harmonic content per frame.

Computed as the maximum of the normalised autocorrelation function over the search range, clipped to [0, 1]. High values indicate periodic (tonal) content; low values indicate noise. Window defaults: WindowLength = round(0.03 · fs), OverlapLength = round(0.02 · fs).

Parameters

x	Real 1-D audio signal.
fs	Sample rate in Hz.
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame harmonic ratios in [0, 1].

See also: pitch

◆ hz2bark()

Value nk::audio::hz2bark	(	const Value &	hz,
		std::pmr::memory_resource *	mr = nullptr )

Hertz → Bark (Traunmüller 1990 with low/high-frequency corrections).

Parameters

hz	Frequency values in Hz.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array.

See also: bark2hz

◆ hz2erb()

Value nk::audio::hz2erb	(	const Value &	hz,
		std::pmr::memory_resource *	mr = nullptr )

Hertz → ERB (Glasberg & Moore 1990).

Equivalent Rectangular Bandwidth scale used in psychoacoustic models.

Parameters

hz	Frequency values in Hz.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array.

See also: erb2hz

◆ hz2mel()

Value nk::audio::hz2mel	(	const Value &	hz,
		std::pmr::memory_resource *	mr = nullptr )

Hertz → Mel (O'Shaughnessy 1987 default form).

\( m = 2595 \log_{10}(1 + f / 700) \)

Parameters

hz	Frequency values in Hz (real array).
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of mel-frequency values.

See also: mel2hz

◆ mel2hz()

Value nk::audio::mel2hz	(	const Value &	mel,
		std::pmr::memory_resource *	mr = nullptr )

Mel → Hertz (inverse of hz2mel).

Parameters

mel	Mel-frequency values.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of Hz values.

◆ melSpectrogram()

std::tuple< Value, Value, Value > nk::audio::melSpectrogram	(	const Value &	x,
		double	fs,
		int	numBands = 32,
		std::pmr::memory_resource *	mr = nullptr )

Mel-band power spectrogram.

Defaults:

NumBands = 32
FrequencyRange = [0, fs/2]
MelStyle = 'oshaughnessy'
Window = hamming(round(0.03·fs), 'periodic')
OverlapLength = round(0.02·fs)
FFTLength = numel(Window)
Normalization = 'bandwidth'
SpectrumType = 'power'
WindowNormalization = true

Parameters

x	Real 1-D audio signal.
fs	Sample rate in Hz.
numBands	Number of mel bands. Default 32.
mr	Memory resource (nullptr → process default).

Returns

Tuple (S, F, T):

S — numBands × NumFrames mel-band power matrix.
F — 1 × numBands band-center frequencies (Hz).
T — NumFrames × 1 frame-center times (seconds).

See also: mfcc, hz2mel, audioDelta

◆ mfcc()

std::tuple< Value, Value, Value > nk::audio::mfcc	(	const Value &	x,
		double	fs,
		int	numCoeffs = 13,
		std::pmr::memory_resource *	mr = nullptr )

Mel-frequency cepstral coefficients (MFCC).

Pipeline:

Hamming (0.03 · fs, 'periodic') STFT.
|FFT| magnitude.
Slaney mel filterbank ('Bandwidth' normalisation).
log10 + DCT-II via cepstralCoefficients.
Natural-log unwindowed-frame energy prepended ('append' default).

Parameters

x	Real 1-D audio signal.
fs	Sample rate in Hz.
numCoeffs	Number of cepstral coefficients. Default 13.
mr	Memory resource (nullptr → process default).

Returns: Tuple (coeffs, delta, deltaDelta) — three matrices, each NumFrames × (numCoeffs + 1). delta and deltaDelta are the first and second time-derivative estimates of the coefficients.

See also: cepstralCoefficients, audioDelta, gtcc

◆ phon2sone()

Value nk::audio::phon2sone	(	const Value &	phon,
		bool	standardIs532_2 = false,
		std::pmr::memory_resource *	mr = nullptr )

Phon → Sone — loudness-level to loudness conversion.

Default standard ISO 532-1: closed-form piecewise power law. Pass standardIs532_2 = true to use ISO 532-2 (table lookup with PCHIP interpolation per Table 5, linear extrapolation beyond 120 phon).

Parameters

phon	Loudness level in phon.
standardIs532_2	Use ISO 532-2 table form instead of 532-1.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of loudness in sone.

See also: sone2phon

◆ pitch()

Value nk::audio::pitch	(	const Value &	x,
		double	fs,
		double	minF = 50.0,
		double	maxF = 400.0,
		std::pmr::memory_resource *	mr = nullptr )

Pitch via the Normalised Cross-correlation Function (the default estimator).

Per-frame f0 estimation via normalised autocorrelation, peak-picked within [minF, maxF]. Window defaults:

WindowLength = round(0.052 · fs)
OverlapLength = round(0.042 · fs)
Range = [50, 400] Hz

Parameters

x	Real 1-D audio signal.
fs	Sample rate in Hz.
minF	Minimum pitch search bound in Hz (default 50).
maxF	Maximum pitch search bound in Hz (default 400).
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame f0 in Hz.

See also: pitchCEP, pitchPEF, pitchLHS, pitchSRH, harmonicRatio

◆ pitchCEP()

Value nk::audio::pitchCEP	(	const Value &	x,
		double	fs,
		double	minF = 50.0,
		double	maxF = 400.0,
		std::pmr::memory_resource *	mr = nullptr )

Pitch via the cepstrum method (Noll 1967).

Same arg semantics as pitch.

Parameters

x	See pitch.
fs	See pitch.
minF	See pitch.
maxF	See pitch.
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame f0 in Hz.

See also: pitch

◆ pitchLHS()

Value nk::audio::pitchLHS	(	const Value &	x,
		double	fs,
		double	minF = 50.0,
		double	maxF = 400.0,
		std::pmr::memory_resource *	mr = nullptr )

Log Harmonic Sum / Subharmonic Summation (Hermes 1988).

Same arg semantics as pitch.

Parameters

x	See pitch.
fs	See pitch.
minF	See pitch.
maxF	See pitch.
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame f0 in Hz.

See also: pitch

◆ pitchPEF()

Value nk::audio::pitchPEF	(	const Value &	x,
		double	fs,
		double	minF = 50.0,
		double	maxF = 400.0,
		std::pmr::memory_resource *	mr = nullptr )

Pitch Estimation Filter (Gonzalez & Brookes 2011).

Same arg semantics as pitch.

Parameters

x	See pitch.
fs	See pitch.
minF	See pitch.
maxF	See pitch.
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame f0 in Hz.

See also: pitch

◆ pitchSRH()

Value nk::audio::pitchSRH	(	const Value &	x,
		double	fs,
		double	minF = 50.0,
		double	maxF = 400.0,
		std::pmr::memory_resource *	mr = nullptr )

Summation of Residual Harmonics (Drugman & Alwan 2011).

LPC inverse filtering + harmonic summation on the residual spectrum. Same arg semantics as pitch.

Parameters

x	See pitch.
fs	See pitch.
minF	See pitch.
maxF	See pitch.
mr	Memory resource (nullptr → process default).

Returns: Column vector of per-frame f0 in Hz.

See also: pitch

◆ sone2phon()

Value nk::audio::sone2phon	(	const Value &	sone,
		bool	standardIs532_2 = false,
		std::pmr::memory_resource *	mr = nullptr )

Sone → Phon — loudness to loudness-level conversion.

Inverse of phon2sone.

Parameters

sone	Loudness in sone (any shape).
standardIs532_2	Use ISO 532-2 table form instead of 532-1.
mr	Memory resource (nullptr → process default).

Returns: Same-shape DOUBLE array of loudness in phon.

See also: phon2sone

◆ spectralCentroid()

Value nk::audio::spectralCentroid	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral centroid — power-weighted mean frequency.

\( \mu_X = \sum_k f_k \cdot X_k \big/ \sum_k X_k \)

Parameters

x	Time signal or M × N spectrum (see file note).
f	Sample rate or frequency vector (see file note).
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column centroid.

See also: spectralSpread

◆ spectralCrest()

Value nk::audio::spectralCrest	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral crest — max(X) / mean(X) per frame.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column crest.

See also: spectralFlatness

◆ spectralDecrease()

Value nk::audio::spectralDecrease	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral decrease — slope-like measure relative to first bin.

\( d = \dfrac{1}{\sum_{k \ge 2} X_k}\,\sum_{k \ge 2} \dfrac{X_k - X_1}{k - 1} \)

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column decrease.

◆ spectralEntropy()

Value nk::audio::spectralEntropy	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral entropy — Shannon entropy of normalised spectrum.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column normalised spectral entropy, in [0, 1].

◆ spectralFlatness()

Value nk::audio::spectralFlatness	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral flatness — geomean(X) / mean(X).

1 = white, 0 = tonal.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column flatness.

See also: spectralCrest

◆ spectralFlux()

Value nk::audio::spectralFlux	(	const Value &	x,
		const Value &	f,
		double	p = 2.0,
		std::pmr::memory_resource *	mr = nullptr )

Spectral flux — p-norm of frame-to-frame spectrum differences.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
p	Norm order (default 2 — Euclidean).
mr	Memory resource (nullptr → process default).

Returns: N - 1 flux values (one per frame transition).

◆ spectralKurtosis()

Value nk::audio::spectralKurtosis	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral kurtosis — X-weighted 4th central moment.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column kurtosis.

See also: spectralSkewness, spectralSpread

◆ spectralRolloffPoint()

Value nk::audio::spectralRolloffPoint	(	const Value &	x,
		const Value &	f,
		double	percentile = 0.95,
		std::pmr::memory_resource *	mr = nullptr )

Spectral rolloff frequency.

f such that the cumulative spectral power below f equals percentile of the total power.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
percentile	Cumulative power fraction in (0, 1). Default 0.95.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column rolloff frequency.

◆ spectralSkewness()

Value nk::audio::spectralSkewness	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral skewness — X-weighted 3rd central moment.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column skewness.

See also: spectralKurtosis, spectralSpread

◆ spectralSlope()

Value nk::audio::spectralSlope	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral slope — least-squares slope of (f, X).

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column slope.

◆ spectralSpread()

Value nk::audio::spectralSpread	(	const Value &	x,
		const Value &	f,
		std::pmr::memory_resource *	mr = nullptr )

Spectral spread — power-weighted std around the centroid.

Parameters

x	Time signal or spectrum.
f	Sample rate or frequency vector.
mr	Memory resource (nullptr → process default).

Returns: Per-frame / per-column spread.

See also: spectralCentroid, spectralSkewness, spectralKurtosis

Files

Classes

Functions

Detailed Description

Function Documentation

◆ audioDelta()

◆ audioinfo()

◆ audioinfoFromBytes() [1/2]

◆ audioinfoFromBytes() [2/2]

◆ audioread()

◆ audioreadFromBytes() [1/2]

◆ audioreadFromBytes() [2/2]

◆ audiowrite()

◆ audiowriteToBytes()

◆ bark2hz()

◆ cepstralCoefficients()

◆ erb2hz()

◆ gtcc()

◆ harmonicRatio()

◆ hz2bark()

◆ hz2erb()

◆ hz2mel()

◆ mel2hz()

◆ melSpectrogram()

◆ mfcc()

◆ phon2sone()

◆ pitch()

◆ pitchCEP()

◆ pitchLHS()

◆ pitchPEF()

◆ pitchSRH()

◆ sone2phon()

◆ spectralCentroid()

◆ spectralCrest()

◆ spectralDecrease()

◆ spectralEntropy()

◆ spectralFlatness()

◆ spectralFlux()

◆ spectralKurtosis()

◆ spectralRolloffPoint()

◆ spectralSkewness()

◆ spectralSlope()

◆ spectralSpread()