synsrc - Synthesis Source function

Synopsis 

synsrc	[-gain:[+-]<db>]
	[-model:R  [-du:<percent>] [-ct:<percent>] [-flow]]
	[-model:LF [-te:<percent>] [-ta:<percent>] [-tp:<percent>]]
	<input.ksp> <output.wav>

Description

synsrc reads a .ksp file containing Klatt synthesis parameters (typically created by program gensyn) and writes a two-channel waveform file. The output waveform file is an excitation signal estimated from the klatt parameters with the voicing signal and aspiration noise on channel 0 and the frication noise on channel 1. This waveform file may in turn be used as the input to program sf (along with the .ksp file from which it was generated) to create a synthetic speech waveform file. Thus, use of the programs synsrc and sf is a substitute for use of the program klatlk which also generates synthetic speech from a .ksp file. The main difference between these two methods is that klatlk uses the source function generated by the original Klatt (1980) synthesis program, whereas, synsrc allows choice of the glottal source model and adjustment of the model parameters. Both of the source models in synsrc tend to sound more natural than the original Klatt source model.

Options

     -gain:[+-]<db>
               Provides overall source amplitude scaling.

     -model:R|LF
               Specify which source model to  use.  R  means  the
               Rosenberg (1970) model; LF is the Liljencrants and
               Fant (1985) model. The LF model  is  the  default.
               Depending  on  which  model is used, several addi-
               tional parameters may be specified.

     -te:<proportion>
               In the LF model,  the  point  of  maximum  closure
               velocity  expressed  as  a proportion of the total
               period. (Default = 0.6)

     -tp:<proportion>
               In the LF model, the point of zero  closure  velo-
               city  expressed  as  a  proportion  of  the  total
               period. (Default = 0.4)

     -ta:<proportion>
               In the LF model, the time from te to a  return  to
               zero   flow  expressed  proportion  of  the  total
               period. (Default = 0.4)

     -du:<proportion>
               In the Rosenberg model, the duty cycle (proportion
               of  the  pitch period during which the vocal folds
               are closed) expressed as a proportion of the total
               period. (Default = 0.6)

     -ct:<proportion>
               In the Rosenberg model, the duration of the  clos-
               ing  portion  of  the glottal cycle expressed as a
               proportion of the total period. (Default = 0.4)

     <input.ksp>
               A binary Klatt Synthesis Parameter  file  as  gen-
               erated by GENSYN.

     <output.wav>
               A two-channel waveform file in RIFF format with chan-
               nel 0 being the voicing source plus aspiration noise
               and channel 1 being the frication noise.