Cepstral Announces Version 4.0 of Cepstral Swift Text-To-Speech System

PITTSBURGH, PA -- (MARKET WIRE) -- 01/31/2006 -- Cepstral LLC, the
synthetic voice company, announced today the release of the award-winning
Cepstral Swift Text-To-Speech (TTS) System, version 4.0, with new voices
and key improvements.

Cepstral Swift Text-To-Speech System 4.0 is a scalable, multilingual
cross-platform voice rendering engine, available at store.cepstral.com, or
through affiliates, partners and resellers worldwide. Voices perform with
low latency and minimum resource load on handhelds, desktops, and large
automated telephony or VoIP installations.

Featured in the release is an engaging new U.S. English female voice,
Callie, well-suited for both telephony and mobile applications, as well as
new novelty voices such as Cepstral Dog and Whispery. Pronunciations and
text processing have been made more accurate for all languages, including
U.S. and U.K. English, German, Canadian French, Americas Spanish, and
Italian.

Among the enhancements are more powerful dictionary features, more capable
run-time behaviour configuration, enhanced mark-up language support, and
changes to the core algorithms that improve the voice experience overall.
64-bit support is available, the Mac OS voices are compatible with the new
Intel-based Apple Macintosh computers; Sparc and x86 Solaris are now both
available, as well as all Windows and Windows Mobile platforms.

Cepstral voices remain competitively priced, starting at $29.95 per voice.
Concurrency licenses, SDKs, and other products make the voices easy to
integrate with low cost, and the Cepstral Affiliate and Reseller programs
allow for affordable self-service wherever spoken output may be used.

About Cepstral LLC

Cepstral is a speech technology company based in Pittsburgh, PA, USA, which
provides customizable Text-To-Speech (TTS) voices and technologies for
server, desktop and hand-held platforms on any operating system. Cepstral:
We Build Voices.