SPEECH RECOGNITION IN A MOTOR VEHICLE
1 Assignment
0 Petitions
Accused Products
Abstract
In a method for speech recognition in a motor vehicle, speech inputs from a user are received and at least one of them is supplied to a vehicle-internal onboard speech recognition system without raising a query with the user. The speech recognition system generates a first recognition result, which, however, is not necessarily always correct and thus it has often been necessary up to now to raise a query with the user or allow the user to make corrections. The method provides reliable speech recognition that distracts the user in the motor vehicle as little as possible from the traffic. To this end, a processing unit supplies all or part of the speech input also to a vehicle-external offboard speech recognition system, which transmits a second recognition result to the processing unit. Depending on the first and the second recognition results, the speech text is then determined.
18 Citations
28 Claims
-
1-13. -13. (canceled)
-
14. A method for speech recognition in a motor vehicle, comprising:
-
receiving voice inputs from a user at a vehicle-internal processing device, the voice inputs being received via a vehicle-internal capture device; supplying by the processing device; at least one voice input to a vehicle-internal onboard speech recognition unit for producing a first recognition result of spoken text; at least one voice input to a vehicle-external offboard speech recognition unit for producing a second recognition result of spoken text; receiving the first recognition result at the processing device from the onboard speech recognition unit; receiving the second recognition result at the processing device from the offboard speech recognition unit; and using the first and second recognition results by the processing device as a basis for automatically ascertaining recognition text corresponding to the voice inputs, wherein the processing device selects and activates based on a control parameter, one of the following modes for speech recognition purposes; sequential speech recognition in which keyword recognition is first performed using the onboard speech recognition unit, and then the first recognition result is used to extract at least one unrecognizable component, which is then supplied to the offboard speech recognition unit to produce the second recognition result; parallel speech recognition in which a single voice input is supplied in parallel to the onboard speech recognition unit and independently thereof to the offboard speech recognition unit; complete onboard speech recognition using the onboard speech recognition unit; and complete offboard speech recognition using the offboard speech recognition unit, and wherein the control parameter on which the processing device bases selection and activation, is a parameter relating to a current operational scenario, the operational scenario indicating what operator control process is meant to be performed by the user by way of the recognition text. - View Dependent Claims (15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26)
-
-
27. An infotainment system for a motor vehicle comprising:
-
a processing device to; receive voice inputs from a user, the voice inputs being received via a vehicle-internal capture device; supply; at least one voice input to a vehicle-internal onboard speech recognition unit to produce a first recognition result of spoken text; at least one voice input to a vehicle-external offboard speech recognition unit to produce a second recognition result of spoken text; receive the first recognition result from the onboard speech recognition unit; receive the second recognition result from the offboard speech recognition unit; and use the first and second recognition results as a basis for automatically ascertaining recognition text corresponding to the voice inputs, wherein the processing device selects and activates based on a control parameter, one of the following modes for speech recognition purposes; sequential speech recognition in which keyword recognition is first performed using the onboard speech recognition unit, and then the first recognition result is used to extract at least one unrecognizable component, which is then supplied to the offboard speech recognition unit to produce the second recognition result; parallel speech recognition in which a single voice input is supplied in parallel to the onboard speech recognition unit and independently thereof to the offboard speech recognition unit; complete onboard speech recognition using the onboard speech recognition unit; and complete offboard speech recognition using the offboard speech recognition unit, and wherein the control parameter on which the processing device bases selection and activation, is a parameter relating to a current operational scenario, the operational scenario indicating what operator control process is meant to be performed by the user by way of the recognition text.
-
-
28. A motor vehicle comprising:
-
an onboard speech recognition unit; a capture device to capture voice inputs from a user and to output a captured voice input; a processing device to; receive the captured voice input from the capture device; supply; at least one voice input to the onboard speech recognition unit to produce a first recognition result of spoken text; at least one voice input to a vehicle-external offboard speech recognition unit to produce a second recognition result of spoken text; receive the first recognition result from the onboard speech recognition unit; receive the second recognition result from the offboard speech recognition unit; and use the first and second recognition results as a basis for automatically ascertaining recognition text corresponding to the voice inputs, wherein the processing device selects and activates based on a control parameter, one of the following modes for speech recognition purposes; sequential speech recognition in which keyword recognition is first performed using the onboard speech recognition unit, and then the first recognition result is used to extract at least one unrecognizable component, which is then supplied to the offboard speech recognition unit to produce the second recognition result; parallel speech recognition in which a single voice input is supplied in parallel to the onboard speech recognition unit and independently thereof to the offboard speech recognition unit; complete onboard speech recognition using the onboard speech recognition unit; and complete offboard speech recognition using the offboard speech recognition unit, and wherein the control parameter on which the processing device bases selection and activation, is a parameter relating to a current operational scenario, the operational scenario indicating what operator control process is meant to be performed by the user by way of the recognition text.
-
Specification