KENWOOD Corporation


November 9, 2004
Press Release

<Announcement>

KENWOOD succeeds in developing voice guidance system, which features a very natural and clear voice closely resembling a real human voice, for advanced vehicle-mounted equipment such as car navigation systems:

Japan's first hybrid, voice synthesis software "HBTTS*."


*Trademark pending.


Summary

KENWOOD Corporation (CEO: Haruo Kawahara, headquarters: Hachioji, Tokyo) has developed the first hybrid voice synthesis software in Japan for use in vehicle-mounted equipment such as car navigation systems.


Details of development

Conventional car navigation systems generally provide voice-based route guidance by reciting edited pre-recorded voice segments. However, with such technological innovations as cellular phones, the Internet and communications equipment modules, today's vehicle-mounted devices fulfill various functions not only as navigation systems but also as advanced telematics terminals. For example, such a terminal searches for facilities like restaurants through various networked databases and offers drivers route guidance to the correct location by voice, or allows them to listen to e-mail messages and news reports.

For this reason, demand is becoming strong for such human-machine interfaces that realize guidance functions together with a natural and clear human-like voice. Meanwhile, conventional navigation systems generally use the voice segment-editing method for guiding routes, and the TTS method is used to recite e-mail messages and news reports. The TTS technology generates a voice by processing signals based on information in the waveform database.
Against this backdrop, KENWOOD has developed Japan's first hybrid voice synthesis software, "HBTTS," which is capable of reciting guidance, e-mail and news texts (processed by voice segment-editing and TTS methods), by the same narrator, and produces fluent voice guidance by augmenting new phrases that are not contained in the voice segment database with sound sources in the waveform database.
KENWOOD is considering using the new software in a car navigation system that it is about to launch.


Technical outline of hybrid voice synthesis software "HBTTS" and its features

Hybrid voice synthesis software:

(1) The voice segment-editing method, with which the pre-recorded "voice" is selected from the voice segment database, is used for voice guidance on routes and operations of equipment.
Voice segment database: a database containing recorded voices
(2) As for new route names, crossings and facilities, which are not registered in the voice segment database, the synthetic guiding voice is created by augmenting the "voice" generation component with a waveform database that then combines the augmented voice and voice segment by using the voice segment-editing engine.
Waveform database: a database including voices divided into minimum segments
(3) When reciting e-mail messages and news reports, which are difficult to prerecord as voice segments, all texts are converted into synthesized speech using the waveform database.
(4) The "speaking speed" when reciting texts, integrating the voice segment-editing and TTS methods, can be freely changed using the "speech" speed conversion function.

As mentioned above, the hybrid voice synthesis software "HBTTS" enables various types of digital content in car navigation and telematics systems to be recited with a uniform voice, to provide guidance in a natural human-like voice by optimally blending the prerecorded voices.

Hybrid voice synthesis software: HBTTS
A hybrid solution combining voice segment-editing and TTS methods enables the reciting quality of the entire texts to be improved.
Voice segment and waveform databases are constructed using the voice of the same narrator.
The reciting speed conversion technology can smooth over the difference in vocal speed between voice segments, and change the speed of reproduction.

Hybrid voice synthesis software: HBTTS

Developed with the technical cooperation of Animo Limited, the hybrid voice synthesis software "HBTTS" uses "FineSpeech," a TTS software technology.


Potential application of hybrid voice synthesis software "HBTTS"

Reciting e-mail messages, news reports, traffic information, weather forecasts, etc. by advanced telematics terminals
Emergency notification by voice through public broadcasting systems
Voice interface for robots
At present, KENWOOD has no plan to provide this software to other companies.

Animo Limited
Animo develops and sells software applications and services, related to voice synthesis, voice authentication, audio signal processing technologies, etc. In March 2001, it commercialized "FineSpeech," a high-quality voice synthesis software, which is extensively being used by corporations and regional governments in Japan in the form of PC applications for audio response and reciting of e-mail messages. It is also used for voice functions of robots.

For queries regarding this article, please contact:

Public Relations Office
KENWOOD Corporation (Japan)
Tel: +81-426-46-6724
Fax: +81-426-46-1440
Email: pr.qa@pr.kenwood.co.jp



Updated 04/11/09 (C) 2004 KENWOOD Corporation