NAME¶
hts_engine — HMM-based speech synthesis engine
SYNOPSIS¶
hts_engine [
options] [
infile]
DESCRIPTION¶
This manual page documents briefly the
hts_engine command.
This manual page was written for the
Debian distribution because the
original program does not have a manual page. Instead, it has documentation in
the GNU
Info format; see below.
hts_engine is a program that synthesize speech waveform from HMMs trained
by the HMM-based speech synthesis system (HTS).
OPTIONS¶
A summary of options is included below.
- -td tree
- decision tree files for state duration
- -tm tree
- Show version of program.
- -tf tree
- decision tree files for Log F0
- -tl tree
- decision tree files for low-pass filter
- -md pdf
- model files for state duration
- -mm pdf
- model files for spectrum
- -mf pdf
- model files for Log F0
- -ml pdf
- model files for low-pass filter
- -dm win
- window files for calculation delta of spectrum
- -df win
- window files for calculation delta of Log F0
- -dl win
- window files for calculation delta of low-pass filter
- -od s
- filename of output label with duration
- -om s
- filename of output spectrum
- -of s
- filename of output Log F0
- -ol s
- filename of output low-pass filter
- -or s
- filename of output raw audio (generated speech)
- -ow s
- filename of output wav audio (generated speech)
- -ot s
- filename of output trace information
- -vp
- use phoneme alignment for duration
- -i i f1 .. fi
- enable interpolation & specify number(i),coefficient(f)
[1][1--]
- -s i
- sampling frequency [16000][1--48000]
- -p i
- frame period (point) [80][1--]
- -a f
- all-pass constant [0.42][0.0--1.0]
- -g i
- gamma = -1 / i (if i=0 then gamma=0) [0][0--]
- -b f
- postfiltering coefficient [0.0][-0.8--8.0]
- -l
- regard input as log gain and output linear one (LSP)
- -r f
- speech speed rate [1.0][0.0--10.0]
- -fm f
- add half-tone [0.0][-24.0--24.0]
- -u f
- voiced/unvoiced threshold[0.5][0.0--1.0]
- -em tree
- decision tree files for GV of spectrum
- -ef tree
- decision tree files for GV of Log F0
- -el tree
- decision tree files for GV of low-pass filter
- -cm pdf
- filenames of GV for spectrum
- -cf pdf
- filenames of GV for Log F0
- -cl pdf
- filenames of GV for low-pass filter
- -jm f
- weight of GV for spectrum [1.0][0.0--2.0]
- -jf f
- weight of GV for Log F0 [1.0][0.0--2.0]
- -jl f
- weight of GV for low-pass filter [1.0][0.0--2.0]
- -k tree
- GV switch
- -z i
- audio buffer size [1600][0--48000]
- infile
- label file
option '-d' may be repeated to use multiple delta parameters. generated
spectrum, log F0, and low-pass filter coefficient sequences are saved in
natural endian, binary (float) format.
EXAMPLE¶
If you installed hts-voice-nitech-jp-atr503-m001 in the current directory, the
following command let you make a voice file from input.lab:
% hts_engine -s 48000 -p 240 -a 0.55 \
-td tree-dur.inf -tm tree-mgc.inf -tf tree-lf0.inf \
-tl tree-lpf.inf -md dur.pdf -mm mgc.pdf \
-mf lf0.pdf -ml lpf.pdf -dm mgc.win1 \
-dm mgc.win2 -dm mgc.win3 -df lf0.win1 \
-df lf0.win2 -df lf0.win3 -dl lpf.win1 \
-em tree-gv-mgc.inf -ef tree-gv-lf0.inf -cm gv-mgc.pdf \
-cf gv-lf0.pdf -k gv-switch.inf -ow output.wav \
input.lab
AUTHOR¶
This manual page was written by Koichi Akabe vbkaisetsu@gmail.com for the
Debian system (and may be used by others). Permission is granted to
copy, distribute and/or modify this document under the terms of the GNU
General Public License, Version 2 any later version published by the Free
Software Foundation.
On Debian systems, the complete text of the GNU General Public License can be
found in /usr/share/common-licenses/GPL.
