Functions for use with frame based processing

Collaboration diagram for Functions for use with frame based processing:

Functions
void	sig2coef (EST_Wave &sig, EST_Track &a, EST_String type, float factor=2.0, EST_WindowFunc *wf=EST_Window::creator(DEFAULT_WINDOW_NAME))
void	sigpr_base (EST_Wave &sig, EST_Track &fv, EST_Features &op, const EST_StrList &slist)
void	power (EST_Wave &sig, EST_Track &a, float factor)
void	energy (EST_Wave &sig, EST_Track &a, float factor)
void	fbank (EST_Wave &sig, EST_Track &fbank, const float factor, EST_WindowFunc *wf=EST_Window::creator(DEFAULT_WINDOW_NAME), const bool up=false, const bool take_log=true)
void	melcep (EST_Wave &sig, EST_Track &mfcc_track, float factor, int fbank_order, float liftering_parameter, EST_WindowFunc *wf=EST_Window::creator(DEFAULT_WINDOW_NAME), const bool include_c0=false, const bool up=false)

Detailed Description

In the following functions, the input is a EST_Wave waveform, and the output is a (usually multi-channel) EST_Track. The track must be set up appropriately before hand. This means the track must be resized accordingly with the correct numbers of frame and channels.

The positions of the frames are found by examination of the time array in the EST_Track, which must be filled prior to the function call. The usual requirement is for fixed frame analysis, where each analysis frame is, say, 10ms after the previous one.

A common alternative is to perform pitch-synchronous analysis where the time shift is related to the local pitch period.

Function Documentation

void sig2coef	(	EST_Wave &	sig,
		EST_Track &	a,
		EST_String	type,
		float	factor = 2.0,
		EST_WindowFunc *	wf = EST_Window::creator(DEFAULT_WINDOW_NAME)
	)

Produce a single set of coefficients from a waveform. The type of coefficient required is given in the argument type.

Parameters

type

{ Possible types are: lpc: linear predictive coding cep: cepstrum coding from lpc coefficients melcep: Mel scale cepstrum coding via fbank fbank: Mel scale log filterbank analysis lsf: line spectral frequencies ref: Linear prediction reflection coefficients power: f0: srpd algorithm energy: root mean square energy }

The order of the analysis is calculated from the number of channels in fv. The positions of the analysis windows must be given by filling in the track's time array.

This function windows the waveform at the intervals given by the track time array. The length of each window is factor * the local time shift. The windowing function is given by wf.

Parameters

sig	input waveform
fv	{output coefficients. These have been pre-allocated and the number of channels in a indicates the order of the analysis.}
type	the types of coefficients to be produced. "lpc", "cep" etc
factor	{the frame length factor, i.e. the analysis frame length will be this times the local pitch period.}
wf	function for windowing. See Windowing mechanisms

Definition at line 399 of file sigpr_utt.cc.

void sigpr_base	(	EST_Wave &	sig,
		EST_Track &	fv,
		EST_Features &	op,
		const EST_StrList &	slist
	)

Produce multiple coefficients from a waveform by repeated calls to sig2coef.

Parameters

sig	input waveform
fv	output coefficients. These have been pre-allocated and the number of channels in a indicates the order of the analysis.
op	Features structure containing options for analysis order, frame shift etc.
slist	list of types of coefficients required, from the set of possible types that sig2coef can take.

Definition at line 138 of file sigpr_utt.cc.

void power	(	EST_Wave &	sig,
		EST_Track &	a,
		float	factor
	)

Calculate the power for each frame of the waveform.

Parameters

sig	input waveform
a	output power track
factor	the frame length factor, i.e. the analysis frame length will be this times the local pitch period.

Definition at line 422 of file sigpr_utt.cc.

void energy	(	EST_Wave &	sig,
		EST_Track &	a,
		float	factor
	)

Calculate the rms energy for each frame of the waveform.

This function calls sig2energy

Parameters

sig	input waveform
a	output coefficients
factor	optional: the frame length factor, i.e. the analysis frame length will be this times the local pitch period.

Definition at line 445 of file sigpr_utt.cc.

void fbank	(	EST_Wave &	sig,
		EST_Track &	fbank,
		const float	factor,
		EST_WindowFunc *	wf = EST_Window::creator(DEFAULT_WINDOW_NAME),
		const bool	up = false,
		const bool	take_log = true
	)

Mel scale filter bank analysis. The Mel scale triangular filters are computed via an FFT (see fastFFT). This routine is required for Mel cepstral analysis (see melcep). The analysis of each frame is done by sig2fbank.

A typical filter bank analysis for speech recognition might use log energy outputs from 20 filters.

Parameters

sig	input waveform
fbank	the output. The number of filters is determined from the number size of this track.
factor	the frame length factor, i.e. the analysis frame length will be this times the local pitch period
wf	function for windowing. See {Windowing mechanisms}
up	whether the filterbank analysis should use power rather than energy.
take_log	whether to take logs of the filter outputs

See also

sig2fbank

melcep

Definition at line 496 of file sigpr_utt.cc.

void melcep	(	EST_Wave &	sig,
		EST_Track &	mfcc_track,
		float	factor,
		int	fbank_order,
		float	liftering_parameter,
		EST_WindowFunc *	wf = EST_Window::creator(DEFAULT_WINDOW_NAME),
		const bool	include_c0 = false,
		const bool	up = false
	)

Mel scale cepstral analysis via filter bank analysis. Cepstral parameters are computed for each frame of speech. The analysis requires fbank . The cepstral analysis of the filterbank outputs is performed by fbank2melcep .

A typical Mel cepstral coefficient (MFCC) analysis for speech recognition might use 12 cepstral coefficients computed from a 20 channel filterbank.

Parameters

sig	input: waveform
mfcc_track	the output
factor	the frame length factor, i.e. the analysis frame length will be this times the local pitch period
fbank_order	the number of Mel scale filters used for the analysis
liftering_parameter	for filtering in the cepstral domain See fbank2melcep
wf	function for windowing. See Windowing mechanisms
include_c0	whether the zero'th cepstral coefficient is to be included
up	whether the filterbank analysis should use power rather than energy.

See also

fbank

fbank2melcep

Definition at line 540 of file sigpr_utt.cc.