tclassp(n) 0.2 tclassp "Advanced Speech Signal Processing Tools"

NAME

tclassp - Advanced Speech Signal Processing Tools

    TABLE OF CONTENTS
    SYNOPSIS
    DESCRIPTION
    Standard Options
    Package Commands
        acfana
        mhspitch
        rmsana
        zcrana
    Wrapper Commands
        affilter
        diffsig
        f0ana
        forest
        rfcana
        spectrum
    Window functions
    BUGS, IDEAS, FEEDBACK
    COPYRIGHT

SYNOPSIS

DESCRIPTION

This package provides tcl bindings for libassp, the Advanced Speech Signal Processing Library by Michel Scheffers.

The assp (advanced speech signal processor) library aims at providing functionality for handling speech signal files in most common audio formats and for performing analyses common in phonetic science/speech science. These will include the calculation of formants, fundamental frequency, root mean square, auto correlation, a variety of spectral analyses, zero crossing rate, filtering etc.

Standard Options

All commands accept the following standard options

-outDir output directory: Store output in output directory. Default: input directory
-ascii: not implemented
-flush: not impemented

All commands accept the following callback options.

-stop varName: varName should be the fully qualified name of an existing variable with a boolean value (i.e. anything Tcl accepts as boolean). While set to false, tclassp will continue until all files in filelist are processed. If set to true, tclassp will finish processing the current file and exit afterwards.
-counter varName: varName should be the fully qualified name of an existing variable with a numerical value. tclassp will increase this value each time it starts processing the next file in filelist.

Package Commands

acfana

::tclassp::acfana ?-option? filelist

Analysis of short-term autocorrelation function. Analysis is performed on each file in filelist. The output is stored in the input directory by default. The options are:

-begin starttime: Start the analysis at starttime seconds in the input file. Default: begin of file.
-end endtime: End the analysis at endtime segonds in the input file. Default: end of file.
-window window function: Use window function instead of blackman (default). See list of window functions for details.
-winShift duration: Set window shift to duration ms. Default: 5.0
-winSize duration: Set window size to duration ms. Default 25.0
-analysisOrder num: Set analysis order to num. Default: sample rate in kHz + 3
-energyNorm: Use energy normalization.
-lengthNorm: Use length normalization.
-extension ext: Set output extension to ext. Default: .acf

mhspitch

::tclassp::mhspitch ?-option? filelist

Pitch analysis of signals using Michel's/Modified Harmonic Sieve algorithm. Analysis is performed on each file in filelist. The output is stored in the input directory by default. For details, see:

Duifhuis, H., Willems, L.F., and Sluyter, R.J. (1982). "Measurement of pitch in speech: An implementation of Goldstein's theory of pitch perception," J.Acoust.Soc.Am. 71, 1568-1580.

Scheffers, M.T.M. (1983). "Simulation of auditory analysis of pitch: An elaboration on the DWS pitch meter," J.Acoust.Soc.Am. 74, 1716-1725.

Allik, J., Mihkla, M. and Ross, J. (1984). "Comment on 'Measurement of pitch in speech: An implementation of Goldstein's theory of pitch perception'," J.Acoust.Soc.Am. 75, 1855-1857.

-begin starttime: Start the analysis at starttime seconds in the input file. Default: begin of file.
-end endtime: End the analysis at endtime segonds in the input file. Default: end of file.
-winShift duration: Set window shift to duration ms. Default: 5.0
-maxPitch freq: Set maximum pitch to freq Hz. Default: 500.0
-minPitch freq: Set minimum pitch to freq Hz. Default: 50.0 Minimum: 25.0
-plainSpectrum: Use plain rather than masked power spectrum.
-extension ext: Set output extension to ext. Default: .pit

rmsana

::tclassp::rmsana ?-option? filelist

Analysis of short-term Root Mean Square amplitude of the signals in filelist. Per default, the RMS values are expressed in decibel (dB) so that they correspond to the short-term power of the signal. The output is stored in the input directory by default.

-begin starttime: Start the analysis at starttime seconds in the input file. Default: begin of file.
-end endtime: End the analysis at endtime segonds in the input file. Default: end of file.
-window window function: Use window function instead of blackman (default). See list of window functions for details.
-winShift duration: Set window shift to duration ms. Default: 5.0
-winSize duration: Set window size to duration ms. Default 25.0
-linVals: Calculate linear RMS values. Default: values in dB
-extension ext: Set output extension to ext. Default: .rms

zcrana

::tclassp::zcrana ?-option? filelist

Analysis of the averages of the short-term positive and negative zero-crossing rates of the signals in filelist. The output is stored in the input directory by default.

-begin starttime: Start the analysis at starttime seconds in the input file. Default: begin of file.
-end endtime: End the analysis at endtime segonds in the input file. Default: end of file.
-winShift duration: Set window shift to duration ms. Default: 5.0
-winSize duration: Set window size to duration ms. Default 25.0
-extension ext: Set output extension to ext. Default: .zcr

Wrapper Commands

The following cammand are implemented via tcl wrapper scripts which call the respective command line tools delivered by the Emu Speech Database System. Over time, their functionality will be implemented in the new assp library.

affilter

::tclassp::affilter ?-option? filelist

Filters audio signals. The Kaiser-window design method is used to compute the coefficients of a linear-phase FIR filter with unity gain in the pass-band. The cut-off frequencies (-6 dB points) of the filters are in the middle of the transition band.

Analysis is performed on each file in filelist. The output is stored in the input directory by default.

-highPass num: Set high-pass cut-off frequency to num Hz. Default: 0, no high-pass filtering
-lowPass num: Set low-pass cut-off frequency to num Hz. Default: 0, no low-pass filtering
-stopBand num: Set the stop-band attenuation to num dB. Default: 93.0 dB, minimum: 21.0 dB
-transition num: Set the width of the transition band to num Hz. Default: 250.0 Hz
-extension ext: Set output extension to ext. The default (auto) resolves to .hpf for high-pass from -highPass, .lpf for low-pass from -lowPass, .bpf for band-pass from -highPass to -lowPass or .bsf for band-stop between -lowPass and -highPass.