!""
#
$
%
&
'
"()*
'
+
,&
-
),
.
!
/
0110
23
/
2
14
56
7
8
9010
:
;
<
=
4
7
8
2
>
1
?
23565
http://freetts.sourceforge.net
FreeTTS is an open source speech synthesis
system developed entirely in the Java
TM
programming language. FreeTTS is based upon
Flite, a synthesis engine written in the C
programming language developed at Carnegie
Mellon University.
The Java
TM
platform has a stigma of being a poor
performer and has often been shunned as an
environment for developing speech engines. To
better understand this stigma, we developed a
speech synthesis engine entirely in the Java
programming language. We discovered that the
Java platform is an excellent platform for a
synthesis engine and can significantly out perform a
similar native C implementation.
What is FreeTTS? FreeTTS is based upon
Flite, a small runtime speech synthesis engine
developed at Carnegie Mellon University
(cmuflite.org). Flite is derived from the Festival
Speech Synthesis System from the University of
Edinburgh and the FestVox project from Carnegie
Mellon University (festvox.org).
FreeTTS includes:
@
A core synthesis engine
@
Support for five general domain voices: an 8
kHz diphone voice, a 16 kHz diphone voice,
and three MBROLA 16kHz diphone voices
@
Supports a high-quality 16 kHz limited
domain voice
@
Partial support for JSAPI 1.0
@
Extensive API documentation
@
Several demonstration programs
The core synthesis engine of FreeTTS is quite
flexible and easily allows new voices to be
added to the system.
How does FreeTTS Work? There are a
number of steps in the synthesis process. Many
of these steps need to be customized depending
on the locale and the type of synthesis
employed. Speech researchers also need the
ability to plug in new algorithms easily.
FreeTTS provides a general framework for the
synthesis process that allows the various steps
in the process to be customized. Figure 1 shows
the overall architecture of FreeTTS.
Figure 1 – FreeTTS Architecture
Voice Data
Lexicon
CARTS
UnitData
Tokenizer
Utterance Processors
Processing
Thread
Wave Synthesizer
Audio Output
Output
Thread
Speakable
Javax.sound
File
Socket
Audio
Utterance
Queue
FreeTTS
Engine
JSAPI
Text JSML Text
Applications
- 1
- 2
- 3
前往页