You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

57 lines
1.7 KiB

12
The Klatt formant synthesizer
12.1 Overview
The final two modules of the MITalk system (CWTRAN and COEWAYV) simulate
a formant synthesizer (Klatt, 1980). Figure 12-1 shows the interface between these
modules and the hardware which produces the actual speech. Synthesizer control
parameter data are specified every 5 msec by rules contained in the phonetic com-
ponent described in Chapter 11. There are 39 control parameters that specify the
actions of the software synthesizer, of which only 20 are varied as a function of
time.
- LOUDSPEAKER
4.9 kHz
LOW-PASS
FILTER
CWTRAN
COEWAV
D/A
CONVERTER
A A
! !
| i
1
CONTROL WAVEFORM
PARAMETERS SAMPLES
(1 SET/5 ms) (10,000/s
12 BITS/SAMPLE)
Figure 12-1: Interface between synthesizer software and hardware
The Klatt formant synthesizer consists of two logically distinct modules,
CWTRAN and COEWAYV. The first modcle, CWTRAN, accepts control
parameter data such as formant freqhéncies, formant bandwidths, and fundamental
frequency (all specified in Hz), as well as source amplitudes and amplitudes of
each parallel formant (specified in dB) and derives a set of difference-equation
coefficients for each digital formant resonator and a set of linear amplitude coef-
ficients. The second synthesizer subroutine, COEWAYV, accepts as input this coef-
ficient and amplitude array for a 5 msec frame and computes the next 5 msec
chunk of waveform.
Considerable system speed improvement can be obtained by implementing
the final module, COEWAYV, as a hardware digital filter. A TTL implementation
has been constructed (Miranker, 1978). Whether the software subroutine or the
VTM is used, output waveforrn samples are played through a digital-to-analog
converter, analog low-pass filter, and loudspeaker.
123