Advanced Development of Speech Enabled Voice Recognition Enabled Embedded Navigation Systems

This paper will focus on the technology improvements and development effort required to enable human sounding system responses and voice enablement for destination entry for automotive navigation systems. Honda and IBM have been working together since around 1999. The first product to reveal the fruits of this relationship was the 2003 Accord. Voice enablement consists of speech recognition of a Radio/Audio, Climate Control, or Navigation related command, and providing voice responses using a combination of prerecorded messages combined with Text-To-Speech (TTS) generated outputs.

The Accord in 2003 used a combination of high quality prerecorded prompts with occasional use of TTS synthesis. The 2005 Acura RL and Odyssey introduced use of Concatenative (human sounding) TTS. This paper will discuss the trade-offs and design considerations for the use of Concatenative TTS as opposed to Formant TTS technology.

In 2005, the Acura RL and Odyssey navigation systems allowed selection by voice of a street name for the United States market. This was an innovative breakthrough in embedded speech technology usage due to the number of street names that at any one time could be stated by the user. This paper will describe the embedded speech technology enhancements that where required for this support and discuss the design and system consideration for destination selection by voice.