Multiple speech recognition engines

• US 20050240404A1
• Filed: 04/23/2004
• Published: 10/27/2005
• Est. Priority Date: 04/23/2004
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: Gurram, Rama, James, Frances
• Granted Patent: US 7,340,395 B2
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems

Abstract

A system having multiple speech recognition engines, each operable to recognize spoken data, is described. A speech recognition engine manager detects the speech recognition engines, and selects at least one for recognizing spoken input from a user, via a user interface. In this way, a speech recognition engine that is particularly suited to a current environment may be selected. For example, a speech recognition engine that is particularly suited for, or preferred by, the user may be selected, or a speech recognition engine that is particularly suited for a particular type of interface, interface element, or application, may be selected. Multiple ones of the speech recognition engines may be selected and simultaneously maintained in an active state, by maintaining a session associated with each of the engines. Accordingly, users'"'"' experience with voice applications may be enhanced, and, in particular, users with physical disabilities may more easily interact with software applications.

Claims

1. A system comprising:

a first speech recognition engine operable to recognize spoken data;

a second speech recognition engine operable to recognize spoken data;

a user interface for receiving spoken input from a user; and

a speech recognition engine manager including a speech recognition engine selector operable to select one of the first speech recognition engine and the second speech recognition engine for recognizing the spoken input from the user.

2. The system of claim 1 wherein the speech recognition engine manager further comprises a speech recognition engine detector operable to detect available speech recognition engines, including the first speech recognition engine and the second speech recognition engine.

3. The system of claim 1 wherein the speech recognition engine selector is operable to select one of the first speech recognition engine and the second speech recognition engine, based on selection information.

4. The system of claim 3 wherein the selection information includes a preference of the user with respect to the first speech recognition engine and the second speech recognition engine.

5. The system of claim 3 wherein the selection information includes predetermined heuristics associated with an operation of the first speech recognition engine and the second speech recognition engine.

6. The system of claim 5 wherein the predetermined heuristics relate to a type of speech recognition associated with the first speech recognition engine or the second speech recognition engine.

7. The system of claim 5 wherein the predetermined heuristics relate to misrecognitions associated with the first speech recognition engine or the second speech recognition engine.

8. The system of claim 3 wherein the selection information includes information related to the user interface with respect to the first speech recognition engine and the second speech recognition engine.

9. The system of claim 1 wherein the speech recognition engine manager uses application program interface (API) libraries to interact with the available speech recognition engines.

10. The system of claim 1 wherein the speech recognition engine manager includes a session manager operable to manage a first session with the first speech recognition engine and a second session with the second speech recognition engine, where the first and second session overlap in time.

11. A method comprising:

determining available speech recognition engines operable to interpret spoken data received in association with a user interface;

selecting a chosen one of the available speech recognition engines to receive spoken input from a user, via the user interface; and

using the chosen speech recognition engine to interpret the spoken input from the user.

12. The method of claim 11 wherein selecting the chosen speech recognition engine includes accessing selection information associated with the user, the user interface, or the available speech recognition engines.

13. The method of claim 11 further comprising:

selecting a plurality of the available speech recognition engines; and

maintaining a session for each of the selected plurality of available speech recognition engines.

14. The method of claim 11 comprising forwarding the interpreted spoken input to the user interface via a voice-enabled portal, so that the interpreted spoken input is displayed as text in association with the user interface.

15. A speech recognition engine manager comprising:

a speech recognition engine detector operable to determine available speech recognition engines; and

a speech recognition engine selector operable to select one of the available speech recognition engines for receiving spoken input from a user of a user interface, based on selection information associated with the user, the user interface, or the available speech recognition engines.

16. The speech recognition engine manager of claim 15 comprising a session manager operable to manage overlapping sessions between the user and at least two of the available speech recognition engines.

17. The speech recognition engine manager of claim 15 wherein the session manager is operable to switch between a first session and a second session, based on a current state of the selection information.

18. The speech recognition engine manager of claim 15 wherein the selection information includes preference information associated with the user with respect to the available speech recognition engines.

19. The speech recognition engine manager of claim 15 wherein the selection information includes predetermined heuristics associated with the user interface or the available speech recognition engines.

20. The system of claim 19 wherein the predetermined heuristics relate to previous misrecognitions by the available speech recognition engines identified with respect to the user.