The authors developed an automatic speech recognition (ASR) system and a text-to-speech system for automotive applications. ASR enhances automotive safety and convenience, and is widely used in many automobile navigation systems, especially in Japan. The described Toshiba's ASR is composed of four parts: a noise reduction part, a voice detection and feature extraction part, a speech recognition part, and a grammar construction part, and then converted into four vectors. At the same time, the system detects the start and end points of the speech. A major feature of the ASR system is high performance in noisy environments. The system is composed of a language-independent and a language-dependent part. This enables to very easily change the language to be recognized and also makes it very easy to construct a multilingual speech recognition system. The language-dependent processing is implemented in the language-dependent part. The performance of the ASR system was evaluated for six European languages. TTS can convert any text into synthesized speech. Toshiba's TTS middleware is widely used for embedded systems such as automobile navigation systems. The TTS system is described in the article. The systems described can operate with various operating systems such as WinCE, VxWorks, Linux and pSOS. They are also small enough to be used in embedded systems such as automobile navigation systems.
Development of multi-lingual speech recognition and text-to-speech synthesis for automotive applications
Entwicklung von multi-lingualen Spracherkennungs- und Sprachausgabesystemen für Anwendungen im Automobil
2002
8 Seiten, 3 Bilder, 4 Tabellen, 5 Quellen
Conference paper
English
British Library Conference Proceedings | 2002
|British Library Conference Proceedings | 2003
|British Library Conference Proceedings | 2002
|Automotive engineering | 2003
|