Elan Speech Launches Elan Sayso for Embedded Platforms

HANNOVER, GERMANY - Elan Speech, a developer of text-to-speech technology, announced the launch of Elan Sayso Embedded. Elan Sayso Embedded technology is able to read text out loud, and uses 10-16MB of memory (depending on the language) for a voice sampled at 22kHz. It therefore overcomes the problem often encountered with embedded platforms, the high memory cost of limited storage capacity and low CPU power. Unit-selection speech synthesis is built from an acoustic base compilation of between 3 to 10 hours of speech. Elan Sayso itself, Elan's unit-selection TTS technology, was market-launched in 2002 following a two-year R&D campaign by Elan Speech engineers. The challenge in developing Elan Sayso Embedded was to sufficiently compact Elan Sayso, while retaining natural tone. Elan's R&D team had to concentrate on finding a way to reduce the sheer size of the acoustic base. Still an ongoing program for Elan's R&D staff, the company's next goal is to reduce footprint to less than 8MB. This should soon be possible in cases where slightly reduced voice quality is sufficient, or where a specialized vocabulary is used in specific applications. Another of Elan's R&D objectives was to develop special linguistic processing tools associated with this technology. These tools render automatic most of the processing needed to create new voices and new languages, and thus guarantee availability of a multilingual product offering, speech-enabled by Elan Sayso. Antoine Kauffeisen, CEO of Elan Speech explains: "Elan Sayso Embedded technology is particularly suited to speech-enabling the automotive world and communicating terminals, such as PDAs and SmartPhones. We are delighted to announce that it is now ready and available to open up new business opportunities."
SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues