Wireless headset and method for robust voice data communication

• US 8,417,185 B2
• Filed: 12/16/2005
• Issued: 04/09/2013
• Est. Priority Date: 12/16/2005
• Status: Active Grant

• Current Assignee: Vocollect, Inc. (Honeywell International Inc.)
• Original Assignee: Vocollect, Inc. (Honeywell International Inc.)
• Inventors: Braho, Keith, Byford, Roger Graham, El-Jaroudi, Amro, Kerr, Thomas S.
• Primary Examiner(s): Lim, Steven
• Assistant Examiner(s): HERRERA, DIEGO D
• Application Number:
  US11/303,271
• Publication Number:
  US 20070143105A1
• Time in Patent Office: 2,671 Days
• Field of Search: 704/232, 455/463, 455/41.2
• US Class Current: 455/41.2
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

A wireless device for use with speech recognition applications comprises a frame generator for generating successive frames from digitized original audio signals, the frames representing portions of the digitized audio signals. An autocorrelation circuit generates a set of coefficients for each frame, the coefficient set being reflective of spectral characteristics of the audio signal portion represented by the frame. In one embodiment, the autocorrelation coefficients may be used to predict the original audio signal to be subtracted from the original audio signals and to generate residual signals A Bluetooth transceiver is configured for transmitting the set of coefficients and/or residual signals as data to another device, which utilizes the coefficients for speech applications.

Claims

1. A wireless device for use with a separate processing device that runs speech recognition applications, the wireless device comprising:

speech processing circuitry for providing initial speech processing of audio captured from a user and including a frame generator for generating successive frames from digitized audio signals that are reflective of the audio signals captured from the user, the frames representing portions of the digitized audio signals; and

autocorrelation circuitry for generating a set of coefficients for each frame, the coefficient set containing information about spectral characteristics of the audio signal portion represented by the frame;

processing circuitry configured for transforming the set of coefficients into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

a Bluetooth transceiver configured for transmitting the set of coefficients as data in the Bluetooth data packet type and configured for using an asynchronous connection-less (ACL) link with a separate processing device for transmitting the set of coefficients to the processing device;

wherein the separate processing device utilizes the coefficients for further processing for speech recognition applications.

2. The wireless device of claim 1, the speech processing circuitry including an A/D circuit for sampling an audio signal captured by the headset to form the digitized audio signals.

3. The wireless device of claim 2 wherein the A/D circuit is configured for sampling at a rate in the range of 11 KHz to 48 KHz.

4. The wireless device of claim 1, the speech processing circuitry including pre-emphasis circuitry to emphasize higher frequency components of the digitized audio signals.

5. The wireless device of claim 4 wherein the pre-emphasis circuitry includes a filter.

6. The wireless device of claim 1 wherein the frame generator generates successive overlapping frames.

7. The wireless device of claim 6 wherein the frames overlap with successive frames in an amount in the range of 0% to 50%.

8. The wireless device of claim 1 wherein the frames generated are in the range of 10 ms to 100 ms in size.

9. The wireless device of claim 1, the speech processing circuitry including windowing circuitry for adjusting the digitized audio signals at the edges of the successive frames.

10. The wireless device of claim 1 wherein the autocorrelation circuitry generates a number of coefficients for each frame in the range of 10 to 21.

11. The wireless device of claim 1 wherein generated coefficients have a data size in the range or 16 to 32 bits per coefficient.

12. A wireless device for use with a separate processing device that runs speech recognition applications, the wireless device comprising:

digital signal processing circuitry for providing initial speech processing and configured to generate sets of coefficients from digitized audio signals that are reflective of audio signals captured from the user, the coefficients containing information about spectral characteristics of portions of the digitized audio signals;

processing circuitry for transforming the sets of coefficients into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

a Bluetooth transceiver configured for transmitting the sets of coefficients as data in the Bluetooth data packet type and configured for using an asynchronous connection-less (ACL) link with a separate processing device for transmitting the set of coefficients to the processing device;

wherein the separate processing device utilizes the coefficients for further processing for speech recognition applications.

13. The wireless device of claim 12 wherein the digital signal processing circuitry is configured for generating successive frames from digitized audio signals, the frames representing portions of the digitized audio signal.

14. The wireless device of claim 13 wherein the frames generated are in the range of 10 ms to 100 ms in size.

15. The wireless device of claim 12 wherein the digital signal processing circuitry pre-emphasizes higher frequency components of the digitized audio signals prior to generating the sets of coefficients.

16. The wireless device of claim 13 wherein the digital signal processing circuitry is configured to perform a windowing application on the successive frames of the digitized audio signals for adjusting the digitized audio signals at the edges of the successive frames.

17. The wireless device of claim 13 wherein the digital signal processing circuitry generates a number of coefficients for each frame in the range of 10 to 21.

18. The wireless device of claim 12 wherein the generated coefficients have a data size in the range or 16 to 32 bits per coefficient.

19. A wireless headset for use with a processing device that runs speech applications, the headset comprising:

a microphone to capture audio signals from a user;

digital signal processing circuitry for providing initial speech processing and configured to generate sets of coefficients from digitized audio signals that are reflective of the audio signals captured from the user, the coefficients containing information about spectral characteristics of portions of the digitized audio signals;

processing circuitry for transforming the set of coefficients into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

a Bluetooth transceiver configured for transmitting the sets of coefficients in the Bluetooth data packet type and configured for using an asynchronous connection-less (ACL) link with a separate processing device for transmitting the set of coefficients to the processing device;

wherein the separate processing device utilizes the coefficients for further processing for speech applications.

20. The wireless headset of claim 19 wherein the digital signal processing circuitry is configured for generating successive frames from digitized audio signals, the frames representing portions of the digitized audio signals.

21. The wireless headset of claim 20 wherein the frames generated are in the range of 10 ms to 100 ms in size.

22. The wireless headset of claim 19 wherein the digital signal processing circuitry pre-emphasizes higher frequency components of the digitized audio signals prior to generating the sets of coefficients.

23. The wireless headset of claim 20 wherein the digital signal processing circuit is configured to perform a windowing application on the successive frames of the digitized audio signal for adjusting the digitized audio signal at the edges of the successive frames.

24. The wireless headset of claim 20 wherein the digital signal processing circuitry generates a number of coefficients for each frame in the range of 10 to 21.

25. The wireless headset of claim 19 wherein the generated coefficients have a data size in the range or 16 to 32 bits per coefficient.

26. A method of transmitting voice information over a wireless link from one device to a separate processing device for use in speech recognition applications, the method comprising:

capturing audio signals from a user with the one device and providing initial speech processing by digitizing the audio signals; and

generating a set of autocorrelation coefficients from the digitized audio signals, the coefficient set containing information about spectral characteristics of the audio signals;

processing the coefficients for transforming the set of coefficients into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

with a Bluetooth transceiver, transmitting the set of coefficients as data in the Bluetooth data packet type to a separate processing device using an asynchronous connection-less (ACL) link with the separate processing device;

utilizing the coefficients in the separate processing device for further processing for speech recognition applications.

27. The method of claim 26 wherein the initial speech processing further includes generating successive frames from a digitized version of the audio signals, the frames representing portions of the digitized audio signals.

28. The method of claim 27 further comprising generating successive overlapping frames.

29. The method of claim 28 wherein the frames overlap with successive frames in an amount in the range of 0% to 50%.

30. The method of claim 26 wherein the initial speech processing further includes pre-emphasizing higher frequency components of the digitized audio signals prior to generating a set of autocorrelation coefficients.

31. The method of claim 27 wherein the frames generated are in the range of 10 ms to 100 ms in size.

32. The method of claim 27 wherein the initial speech processing further includes performing a windowing operation on the frames for adjusting the digitized audio signals at the edges of the successive frames.

33. The method of claim 27 further comprising generating the autocorrelation coefficients with a number of coefficients for each frame in the range of 10 to 21.

34. The method of claim 26 wherein generated coefficients have a data size in the range or 16 to 32 bits per coefficient.

35. A wireless device for use with a separate processing device that runs speech recognition applications, the wireless device comprising:

speech processing circuitry for providing initial speech processing and including conversion circuitry for digitizing original audio signals to form digitized audio signals; and

autocorrelation circuitry for generating a set of coefficients from the digitized audio signals, the coefficient containing information about spectral characteristics of the audio signals;

processing circuitry using the autocorrelation coefficients for generating a predicted version of the audio signals and to subtract the predicted version from the original audio signals to generate residual signals;

processing circuitry for transforming the set of coefficients and residual signals into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

a Bluetooth transceiver configured for transmitting the set of coefficients and the residual signals as data in the Bluetooth data packet type and configured for using an asynchronous connection-less (ACL) link with a separate processing device for transmitting the set of coefficients and residual signals to the processing device;

wherein the separate processing device utilizes the coefficients and the residual signals data for further processing for speech recognition applications.

36. The wireless device of claim 35 wherein the processing circuitry generates prediction coefficients for the predicted version of the audio signals.

37. The wireless device of claim 35 wherein the processing circuitry is configured for normalizing the residual signals.

38. The wireless device of claim 35 further comprising encoding circuitry for encoding the residual signals for transmission.

39. The wireless device of claim 38 further wherein the encoding circuit encodes the residual signals with 2-10 bits.

40. A wireless system for use with speech applications comprising:

a transmitter including speech processing circuitry for providing initial speech processing, the speech processing circuitry configured for digitizing original audio signals and generating a set of autocorrelation coefficients from the digitized audio signals that contain information about spectral characteristics of the audio signals, the speech processing circuitry further configured for generating predicted versions of the audio signals from the autocorrelation coefficients and for subtracting the predicted versions from the original audio signals to generate residual signals, the processing circuitry configured for transforming the set of coefficients and residual signals into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

a Bluetooth transceiver configured for transmitting the set of autocorrelation coefficients and the residual signals as data in the Bluetooth data packet type and configured for using an asynchronous connection-less (ACL) link with a separate processing device for transmitting the set of coefficients and residual signals to the separate processing device; and

the processing device having a receiver for receiving the data, the processing device configured for utilizing the autocorrelation coefficients and residual signals data for further processing for speech applications.

41. The system of claim 40 wherein the processing device is configured to use the autocorrelation coefficients and residual signals data for speech recognition applications.

42. The system of claim 40 wherein the processing device is configured for using the autocorrelation coefficients and residual signals to recreate audio signals reflective of the original audio signals.

43. The system of claim 42 wherein the processing device is configured for using the autocorrelation coefficients to generate predicted versions of the audio signals and then to add the predicted versions to the residual signals to recreate audio signals reflective of the original audio signals.

44. A method of transmitting voice information over a wireless link from one device to a separate processing device for use in speech applications, the method comprising:

capturing an original audio signal with the one device and providing initial speech processing by digitizing the audio signal; and

generating a set of autocorrelation coefficients from the digitized audio signal, the autocorrelation coefficient set containing information about spectral characteristics of the audio signal;

generating a predicted version of the audio signals with the autocorrelation coefficients and subtracting the predicted version from the original audio signals to generate residual signals;

processing the coefficients and residual signals for transforming them into data in a Bluetooth data packet type instead of a voice packet type for wireless transmission;

with a Bluetooth transceiver, transmitting the set of coefficients and the residual signals as data in the Bluetooth data packet type to a separate processing device using an asynchronous connection-less (ACL) link with the separate processing device where the autocorrelation coefficients and residual signals data are utilized by the processing device for speech applications.