Structure for grammar and dictionary representation in voice recognition and method for simplifying link and node-generated grammars

• US 7,921,011 B2
• Filed: 05/19/2006
• Issued: 04/05/2011
• Est. Priority Date: 05/20/2005
• Status: Active Grant

• Current Assignee: Sony Interactive Entertainment Inc. (Sony Group Corp.)
• Original Assignee: Sony Computer Entertainment Incorporated (Sony Group Corp.)
• Inventors: Chen, Ruxin, Hernandez Abrego, Gustavo
• Primary Examiner(s): AZAD, ABUL K
• Application Number:
  US11/437,444
• Publication Number:
  US 20060277032A1
• Time in Patent Office: 1,782 Days
• Field of Search: 704251-257
• US Class Current: 704/254
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks
• CPC Class Codes: G10L 15/285 Memory allocation or algorithm optimisation to reduce hardware requirements

Abstract

Methods for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event are provided. One method includes receiving a set of phrases, the set of phrases being relevant for the computing event and the set of phrases having a node and link structure. Also included is identifying redundant nodes by examining the node and link structures of each of the set of phrases so as to generate a single node for the redundant nodes. The method further includes examining the node and link structures to identify nodes that are capable of being vertically grouped and grouping the identified nodes to define vertical word groups. The method continues with fusing nodes of the set of phrases that are not vertically grouped into fused word groups. Wherein the vertical word groups and the fused word groups are linked to define an optimized grammar structure. In another aspect, a layered grammar and dictionary library that can be defined for efficient use in speech recognition systems, is provided.

Claims

1. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event, comprising:

(a) receiving the set of phrases, the set of phrases being relevant for the computing event;

(a1) defining each of the set of phrases in a node and link structure, wherein a given node is defined by one or more words, and wherein a given link connects the one or more words of one node to the one or more words of another node;

(b) identifying redundant nodes by examining the node and link structures of each of the set of phrases so as to generate and substitute a single node for the redundant nodes, wherein redundant nodes are defined by identical words, and wherein each of the redundant nodes has a respective outgoing link to a same successor node;

(c) examining the node and link structures to identify nodes that are capable of being vertically grouped and grouping the identified nodes to define vertical word groups; and

(d) fusing horizontally linked nodes of the set of phrases that are not vertically grouped and that represent a single sentence path into fused word groups,

wherein the vertical word groups and the fused word groups are linked to define an optimized grammar structure.

2. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the computing event includes computer assisted voice communication, computer triggered voice gaming commands, computer voice assisted gaming, and computer triggered voice commands.

3. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the set of phrases includes two or more spoken or written words, or groups of words.

4. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the node and link structure of the set of phrases includes a single node or multiple nodes connected by a directional link that indicates a progression of the nodes of a particular one of the set of phrases.

5. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein a plurality of nodes are capable of being vertically grouped when each of the plurality of nodes is linked to both a same predecessor and a same successor node.

6. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein fusing horizontally linked nodes includes generating and substituting a single multi-word node for the fused horizontally linked nodes.

7. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the optimized grammar structure is defined by a set of nodes that commence at a begin node and terminate at an end node, and the begin node is linked through the optimized grammar structure to the end node.

8. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 7, wherein the optimized grammar structure is constrained to only produce the phrases defined by the received set of phrases.

9. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 6, wherein a first node is fused with a second node when,

the first node is horizontally linked to the second node,

each of the first and second nodes has a respective word count of one,

the first node is the only predecessor node to which the second node is horizontally linked,

the second node is the only successor node to which the first node is horizontally linked, and

the first node has only one predecessor node horizontally linked to itself.

10. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein different sets of phrases for different computing events are processed by operations (a)-(d) to generate another optimized grammar structure for the different computing event.

11. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the grammar structure is further optimized by removing word groups based on transition probabilities from n-tuples.

12. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein the grammar structure is further optimized by removing sentences based on transition probabilities from n-tuples.

13. A method for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event as recited in claim 1, wherein an additional step (e) examining the node and link structures to identify nodes that are capable of being vertically grouped and grouping the identified nodes to define vertical word groups;

is added after the step (d).

14. A data storage device for storing data to be read by a computer system including program instructions for optimizing grammar structure for a set of phrases to be used in speech recognition during a computing event, comprising:

(a) program instructions for receiving the set of phrases, the set of phrases being relevant for the computing event;

(a1) program instructions for defining each of the set of phrases in a node and link structure, wherein a given node is defined by one or more words, and wherein a given link connects the one or more words of one node to the one or more words of another node;

(b) program instructions for identifying redundant nodes by examining the node and link structures of each of the set of phrases so as to generate and substitute a single node for the redundant nodes, wherein redundant nodes are defined by identical words, and wherein each of the redundant nodes has a respective outgoing link to a same successor node;

(c) program instructions for examining the node and link structures to identify nodes that are capable of being vertically grouped and grouping the identified nodes to define vertical word groups; and

(d) program instructions for fusing horizontally linked nodes of the set of phrases that are not vertically grouped and that represent a single sentence path into fused word groups,

wherein the vertical word groups and the fused word groups are linked to define an optimized grammar structure.

15. A data storage device for storing data to be read by a computer system including program instructions as recited in claim 14, wherein the computing event includes computer assisted voice communication, computer triggered voice gaming commands, computer voice assisted gaming, and computer triggered voice commands.

16. A data storage device for storing data to be read by a computer system including program instructions as recited in claim 14, wherein a plurality of nodes are capable of being vertically grouped when each of the plurality of nodes is linked to both a same predecessor and a same successor node.

17. A data storage device for storing data to be read by a computer system including program instructions as recited in claim 14, wherein fusing horizontally linked nodes includes generating and substituting a single multi-word node for the fused horizontally linked nodes.

18. A data storage device for storing data to be read by a computer system including program instructions as recited in claim 17, wherein a first node is fused with a second node when,

the first node is horizontally linked to the second node,

each of the first and second nodes has a respective word count of one,

the first node is the only predecessor node to which the second node is horizontally linked,

the second node is the only successor node to which the first node is horizontally linked, and

the first node has only one predecessor node horizontally linked to itself.