SpeechWorks Releases ETI-Eloquence SF

BOSTON, MA - SpeechWorks International Inc. (Nasdaq: SPWX), a developer of speech technologies and services, introduced a new embedded text-to-speech (TTS) engine that offers voice output in 13 languages with a reduced memory footprint of less than 90KB for one language. ETI-Eloquence SF (Small Footprint) additionally offers memory performance of less than 200KB for up to five languages on a single cellular phone or other mobile device. This software is optimized for name and short phrase applications such as voice-activated-dialing and SMS reading. ETI-Eloquence SF software is based on SpeechWorks' formant based TTS engine, ETI-Eloquence software. ETI-Eloquence SF offers the same voice output as ETI-Eloquence but with a fraction of the memory required. The ultra-low memory footprint of ETI-Eloquence SF enables applications like voice dialing and SMS messaging, which were previously not possible on mainstream handsets due to the memory and CPU constraints on these devices. According to Gartner Group, approximately 12 billion SMS text messages are sent worldwide each month to PCs, PDAs and mobile phones. Users are demanding that an increasing number of SMS messages be speech-enabled for hands-free playback or for safety purposes such as in-car review. ETI-Eloquence SF will also make mainstream handsets accessible to visually impaired users by enabling audio output of items on screens, including menu and calendar entries. "The growing popularity of SMS and voice activated dialing among cell phone users around the world has created the need for a smaller footprint TTS engine that sounds highly intelligible," said Alan Schwartz, Vice President of the Automotive & Mobile Device Customer Focus Unit at SpeechWorks. "SpeechWorks is now able to provide handset makers with a TTS engine that delivers exceptional speech performance for thirteen languages within the footprint requirements of today's mainstream handsets." SpeechWorks is able to customize the ETI-Eloquence SF voice for developers by manipulating a set of parameters such as breathiness, roughness, gender, speaking rate and prosody. New voices can be made available to users as a download, similar to ring tone applications. In addition, ETI-Eloquence SF software can be shipped as a complete TTS engine, or in two software modules: text-to-phoneme and phoneme-to-speech. The ability to offer ETI-Eloquence SF as two separate modules is ideal for use in handsets that may already support text-to-phoneme for speech recognition on the device, thus enabling the handset to maintain an even lower memory footprint. ETI-Eloquence SF software offers synthesized speech in a small memory footprint in U.S. English, UK English, Continental French, Castilian Spanish, German, Japanese, Brazilian Portuguese, Mexican Spanish, Canadian French, Finnish, Italian, Korean and Mandarin Chinese.