Single processor programmable speech recognition test system

• US 5,715,369 A
• Filed: 11/27/1995
• Issued: 02/03/1998
• Est. Priority Date: 11/27/1995
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Landon, Gregory G., Wittel, Walter Irving Jr., Rozak, Michael J., Spoltman, James H.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): Dorvil, Richemond
• Application Number:
  US08/562,760
• Time in Patent Office: 799 Days
• Field of Search: 395/2.79, 395/2.4, 395/2.45, 395/2.81, 395/2.5, 395/2.52, 395/2.53, 395/2.55, 395/2.56, 395/2.6, 395/2.61
• US Class Current: 704/270
• CPC Class Codes: G10L 15/01 Assessment or evaluation of speech recognition systems

Abstract

A speech recognition test system comprising a single host processing system having a host processor and a memory device, wherein the memory device contains a plurality of audio files accessible by the host processor. The test system also includes a speech recognition application having a vocabulary, an independent test application, a means for concurrently executing the speech recognition application and the independent test application on the host processor, a means for queuing the audio files as input to the speech recognition application by way of the test application, a means for programming the test application at configuration time to expand the vocabulary of the application being tested and/or other voice gender, volume, and speed playback parameters, and a means for capturing and evaluating test results from the speech recognition application by way of the test application. In an alternative embodiment the speech recognition test system includes an audio input/output device operatively connected to the host processor, and a means for redirecting output from the audio input/output system as input to itself.

Claims

1. A speech recognition test system comprising:

a single host processing system having a host processor and a memory device, wherein said memory device contains a plurality of audio files accessible by said host processor and each of said plurality of audio files having at least one audio phrase therein;

a speech recognition application having a vocabulary;

a test application;

means for concurrently executing said speech recognition application and said test application on said host processor; and

means for exercising said speech recognition by way of said test application in a manner that includes;

means for selectively queuing a set of said plurality of audio files as a transparent input source to said speech recognition application by way of said test application;

means for expanding said vocabulary of said speech recognition application in response to previously unknown phrases input in real time to said speech recognition application by said test application; and

means for capturing output from said speech recognition application.

2. A speech recognition test system according to claim 1 including:

means for evaluating said output relative to a model output baseline.

3. A speech recognition test system according to claim 1 including:

means for controlling playback characteristics including voice gender, volume, and speed, for said set of said plurality of audio files by way of said test application.

4. A speech recognition test system according to claim 1 including:

an audio input/output system operatively connected to said host processor; and

means for redirecting output from said audio input/output system as input to itself.

5. A speech recognition test system according to claim 1 wherein said test application is transparent to said speech recognition application.

6. A speech recognition test system according to claim 1 including:

means for displaying said output from said speech recognition application for user viewing.

7. A method for using a speech recognition test system in a single host processing system wherein a host processor within said single host processing system has access to a memory device that contains at least one audio file each having at least one phrase therein, said method comprising the steps of:

concurrently executing a speech recognition application and a test application on said host processor, wherein said test application is transparent to said speech recognition application;

exercising said speech recognition application by said test application in a manner that includes;

controlling playback characteristics including voice gender, volume, and speed, for said at least one phrase on said at least one audio file;

queuing said at least one audio file for said speech recognition application to receive as input from an input source; and

capturing output from said speech recognition application.

8. A method according to claim 7 wherein said step of capturing includes:

recording said output in a result file;

evaluating said output relative to a model output baseline.

9. A method according to claim 8 including:

displaying an evaluation result of said output on a visual monitor concurrently with a graphic representation of audio file data that was input to the speech recognition application to produce said output, in response to said step of evaluating.

10. A method according to claim 7 wherein said step of queuing includes:

repeating any segment of any one of said at least one audio file on said queue.

11. A method according to claim 7 including:

replaying an amount of white noise between a first of said at least one phrase and a second of said at least one phrase.

12. A method according to claim 7 including:

redirecting output from an audio input/output system as input to itself, wherein said audio input/output system is operatively connected to said host processor.

13. A method according to claim 7 wherein said input source is transparent to said speech recognition application.

14. A method for using a speech recognition test system in a single host processing system wherein a host processor within said single host processing system has access to a memory device that contains at least one audio file each having at least one phrase therein, said method comprising the steps of:

concurrently executing a speech recognition application and an independent test application that is transparent to said speech recognition application on said host processor;

exercising said speech recognition application by way of said independent test application in a manner that includes;

controlling test configuration parameters including voice gender selection, playback volume, and playback speed;

queuing said at least one phrase for input to said speech recognition application from said test application;

controlling audio file input to said speech recognition application in view of said test configuration parameters;

expanding said vocabulary of said speech recognition application in response to previously unknown phrases input in real time to said speech recognition application by said independent test application; and

replaying said at least one phrase to said speech recognition application from said test application by way of a single sound card that is operatively connected to said host processor.

15. A method according to claim 14 including:

evaluating said output relative to a model output baseline.

16. A method according to claim 15 including:

displaying an evaluation of said output on a monitor concurrently with a graphic representation of audio file data being input to said speech recognition application.

17. A method according to claim 14 wherein said input is from a source that is transparent to said speech recognition application.

18. A method according to claim 14 including:

directing audio data as output through said sound card; and

redirecting said output from said sound card as input to said sound card by way of a redirection circuit in said sound card.