Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
2000-06-28
2001-05-08
Ŝmits, T{overscore (a)}livaldis I. (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S270000, C704S231000, CD12S415000, CD12S421000, C454S074000
Reexamination Certificate
active
06230138
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates generally to the field of speech recognition systems. In particular, this invention relates to in-vehicle speech recognition systems for controlling vehicle accessories.
Speech recognition systems on board automobiles permit drivers and passengers to control various vehicle functions by uttering voice commands. One or more microphones placed throughout the cabin and coupled to a speech processor receive the voice commands. The speech processor processes the uttered voice commands using various acoustic and language modeling techniques to correspond the uttered voice commands to known commands contained in one or more command grammar sets.
A major problem of existing speech recognition systems is accurately recognizing utterances of multiple speakers. This problem is especially prevalent in speech recognition systems used on board vehicles where multiple occupants, such as the driver, front passengers and rear passengers, issue voice commands. Thus, in-vehicle speech recognition system must be able to process audio signals corresponding to spoken commands issued from various locations within the cabin, both in the front and back seats. In many vehicles, particularly large automobiles, trucks and vans, the acoustic characteristics of the various seating locations can vary with the effects of traffic, window position and audio system sound. This variation in acoustic characteristics can have a noticeably adverse affect on the accuracy of the speech recognition system. Moreover, while simply placing a microphone proximate each seat may help improve overall recognition accuracy, it does not accommodate for the variation in acoustic characteristics at each seat.
Another problem specific to speech recognition systems in automobiles is selectively determining the vehicle functions that may be voice activated by various occupants of the vehicle. For example, in an automobile equipped with one or more voice activated rearview mirrors, permitting passengers to operate the mirror may be disadvantageous. As other examples, it may be desired that a rear seat passenger be prevented from operating a front window, or that a child passenger operate any window controls at all.
Accordingly, an improved in-vehicle speech recognition system is needed that can better recognize speech commands at various cabin locations as well as control which automobile accessories can be voice operated at these locations.
BRIEF SUMMARY OF THE INVENTION
The present invention provides an in-vehicle speech recognition system capable of accurately recognizing speech from vehicle occupants at different cabin locations. The system of the present invention includes multiple speech engines having different recognition characteristics and/or active grammar sets. This allows the system of the present invention to accommodate different acoustic characteristics of various cabin locations and make available different voice operable accessories and accessory-functions at each cabin location.
In one aspect of the invention, a speech recognition system is provided in an automobile having at least one voice activated accessory and a passenger cabin with a plurality of speaking locations. The system includes a plurality of microphones and push-to-talk controls mounted within the passenger cabin so as to correspond to each of the speaking locations. Each of the push-to-talk controls supply a location signal when operated and each of the microphones can receive a spoken utterance from a speaker and provide a corresponding audio signal. A plurality of speech engines, each having different recognition parameters, recognize the audio signal as a speech command for controlling the relevant voice activated accessory. A selector is coupled to the push-to-talk controls, microphones and speech engines. The selector selects one of the speech engines according to the location signal it receives and provides the corresponding audio signal to the selected speech engine.
In another aspect, the selector may be coupled to the push-to-talk controls via a vehicle network in which case the selector processes the location signal to determine which speech processor to select. Alternatively, the selector may have separate inputs for each push-to-talk control. In this case, each input can be directly coupled to a particular speech engine.
In yet another aspect, the selector provides a selection signal to the selected speech engine. Further, one or more of the speech engines can contain multiple grammar sets, each containing different sets of voice commands. In this case, the selector will select an active grammar from the multiple grammar sets according to the location signal. Moreover, the selection signal sent to the selected speech engine will indicate which grammar set is to be the active grammar set.
In still another aspect, the multiple speech engines have different acoustic modeling parameters for processing audio signals corresponding to spoken utterances originating at speaking locations with differing acoustic characteristics and/or microphone placement.
In another aspect, the selector and speech engines are included in a central control unit having a memory for storing voice commands.
Thus, the present invention provides the advantages of an in-vehicle speech recognition system that can accurately recognize voice commands from automobile occupants seated at various cabin locations by providing multiple speech processors each tuned to recognized speech signals originating from locations having different acoustical qualities. Moreover, the present invention provides the additional advantage of selectively controlling what accessories and accessory functions can be controlled at each cabin location.
These and other advantages of the present invention will be apparent from the description of the preferred embodiments, which follow.
Ŝmits T{overscore (a)}livaldis I.
Nolan Daniel A
Quarles & Brady LLP
Visteon Global Technologies Inc.
LandOfFree
Method and apparatus for controlling multiple speech engines... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for controlling multiple speech engines..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for controlling multiple speech engines... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2475196