Automated speech recognition (ASR) tiling

• US 8,219,399 B2
• Filed: 12/20/2007
• Issued: 07/10/2012
• Est. Priority Date: 07/11/2007
• Status: Active Grant

• Current Assignee: Garmin Limited
• Original Assignee: Garmin Switzerland GmbH (Garmin Limited)
• Inventors: Lutz, Pascal M., Caire, Jacob W., Bolton, Kenneth A., Stricklin, Cory J.
• Primary Examiner(s): Vo, Huyen X.
• Application Number:
  US11/960,965
• Publication Number:
  US 20090018840A1
• Time in Patent Office: 1,664 Days
• Field of Search: 704/275, 704/257, 704/270, 704/231, 704/9, 704/4, 704/10, 704/270.1, 704/251, 704/254, 704/255, 704/1, 704/235, 704/272
• US Class Current: 704/270
• CPC Class Codes: G01C 21/36 Input/output arrangements for on-board computers
• CPC Class Codes: G01C 21/3608 using speech input, e.g. using speech recognition
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Techniques are described related to tiles of automated speech recognition data. In an implementation, automated speech recognition (ASR) data is obtained. The ASR data is divided into a plurality of tiles based on an approximate amount of data to be included in each tile. Each of the tiles is a partition of the ASR data.

Claims

1. A method comprising:

obtaining, by a processor associated with a portable electronic device, automated speech recognition (ASR) data, the automated speech recognition (ASR) data configured to translate an audio input;

dividing, by the processor, the ASR data into a plurality of tiles based on an approximate amount of data to be included in each tile, wherein each of the tiles is a partition of the automated speech recognition (ASR) data and the plurality of tiles include a common tile and a plurality of geographically-partitioned tiles, wherein the common tile includes automated speech recognition (ASR) data common to at least two of the plurality of geographically-partitioned tiles, wherein the geographically-partitioned tiles are sized based on a minimum geographic search area; and

using, by the processor, the common tile and at least one of the geographically-partitioned tiles to translate an audio input received by the portable electronic device,

wherein the dividing includes—

forming one or more common tiles that include a portion of said automated speech recognition (ASR) data that corresponds to particular goods or services, and

forming one or more geographically-partitioned tiles that include a portion of said automated speech recognition (ASR) data.

2. A method as described in claim 1, wherein the ASR data includes geo-coded points of interest and addresses.

3. A method as described in claim 1, wherein the dividing includes forming one or more common tiles that include said automated speech recognition (ASR) data that are likely to be repeated in a plurality of the tiles otherwise.

4. A method as described in claim 1, wherein the dividing includes forming one or more common tiles that include a portion of said automated speech recognition (ASR) data that has been identified as having an increased likelihood of being requested by a user and is relevant to a particular geographic area that corresponds to respective said tiles.

5. A method as described in claim 1, wherein the dividing includes forming one or more common tiles that include a portion of said automated speech recognition (ASR) data that has been identified as being likely to occur multiple times in the ASR data otherwise and is relevant to a particular geographic area that corresponds to respective said tiles.

6. A position-determining device comprising:

a processor; and

memory configured to maintain;

automated speech recognition (ASR) data divided according to partitions into a plurality of tiles, at least one of the tiles being configured as a common tile and a plurality of the tiles being geographically partitioned, wherein the common tile includes automated speech recognition (ASR) data common to at least two of the plurality of geographically-partitioned tiles, wherein the geographically-partitioned tiles are sized based on a minimum geographic search area; and

one or more modules that are executable on the processor to;

select one or more of the plurality of tiles based on information related to a user-initiated search and a geographic position,

select the common tile based on the search information but not the geographic position; and

translate an audio input using the selected tiles.

7. A position-determining device as described in claim 6, wherein at least one of the common tiles includes a portion of said automated speech recognition (ASR) data that corresponds to particular goods or services.

8. A position-determining device as described in claim 6, wherein a plurality of said common tiles is organized into respective categories that are searched based on the information.

9. A position-determining device as described in claim 6, wherein at least one of the common tiles includes a portion of said automated speech recognition (ASR) data that has been identified of having an increased likelihood of being requested by a user and is relevant to the geographic position.

10. A position-determining device as described in claim 6, wherein the one or more modules are further executable on the processor to perform one or more functions using the audio input.

11. A position-determining device as described in claim 6, wherein the ASR titles are partitioned by category.