Method and apparatus for reducing synchronization delay in packet-based voice terminals by resynchronizing during talk spurts

• US 7,319,703 B2
• Filed: 07/02/2002
• Issued: 01/15/2008
• Est. Priority Date: 09/04/2001
• Status: Active Grant

• Current Assignee: III Holdings 3, LLC
• Original Assignee: Nokia Corporation
• Inventors: Ojala, Pasi, Selin, Jari, Lakaniemi, Ari
• Primary Examiner(s): Pham; Chi
• Assistant Examiner(s): Chou; Albert T.
• Application Number:
  US10/189,068
• Publication Number:
  US 20030043856A1
• Time in Patent Office: 2,023 Days
• Field of Search: 370/252, 370/412, 370/503, 370/505, 700/94, 704/220
• US Class Current: 370/505
• CPC Class Codes: H04J 3/0632 Synchronisation of packets and cells, e.g. transmission of voice via a packet network, circuit emulation service [CES] queuing arrangements in packet switching elements H04L49/90; synchronising systems for the synchronous transmission of a pulse code modulated video signal with one or more other pulse code modulated signals H04N7/56
• CPC Class Codes: H04L 12/6418 Hybrid transport
• CPC Class Codes: H04L 2012/6481 Speech, voice

Abstract

Circuitry, embodied in a media subsystem (10A), reproduces a speech or other type of audio signal, and is operable when playing back audio data for reducing synchronization delay. A method operates by, when a frame containing audio data is sent to a decoder (20), measuring the synchronization delay; determining by how much the synchronization delay should be adjusted; and adjusting the synchronization delay in a content-aware manner by adding or removing one or more audio samples in a selected current frame or in a selected subsequent frame so as not to significantly degrade the quality of the played back audio data. When the synchronization delay is adjusted by more than one audio sample, the adjustment can be made by all of the determined audio samples in one adjustment, or the adjustment is made by less than all of the determined audio samples by a plurality of adjustments. The step of adjusting selects, if possible, an unvoiced frame and discriminates against a transient frame. The step of determining includes measuring an average amount of time that a frame resides in a jitter buffer (18), and adjusting the synchronization delay so that the average duration approaches a desired jitter buffer residency duration.

Claims

1. A method comprising:

when a frame comprising active speech data is sent to a decoder, measuring synchronization delay;

determining by how much the synchronization delay should be adjusted;

selecting a frame of the active speech data in dependence on characteristics of the active speech data reflected in the frame; and

adjusting the synchronization delay during a period of active speech by adding or removing one or more audio samples in the selected frame so as not to significantly degrade the quality of the played back active speech data.

2. A method as in claim 1, where the step of determining determines the number of audio samples in steps of size one sample.

3. A method as in claim 1, where when the synchronization delay is adjusted by more than one audio sample, the adjustment is made by all of the determined audio samples in one adjustment.

4. A method as in claim 1, where when the synchronization delay is adjusted by more than one audio sample, the adjustment is made by less than all of the determined audio samples by a plurality of adjustments.

5. A method as in claim 1, where the step of selecting selects, if possible, an unvoiced frame.

6. A method as in claim 1, where the step of selecting discriminates against a transient frame.

7. A method as in claim 1, where the step of determining comprises measuring an average amount of time that a frame resides in a jitter buffer;

and adjusting the synchronization delay so that the average duration approaches a desired jitter buffer residency duration.

8. A method as in claim 1, where at least one frame of audio data has a playback duration in the range of about 20 milliseconds to about 60 milliseconds.

9. Apparatus comprising:

buffer circuitry for storing active speech data from a packet prior to the active speech data being sent to a decoder; and

control circuitry operable when a frame comprising active speech data is sent to the decoder, the control circuitry configured to measure synchronization delay;

to determine by how much the synchronization delay should be adjusted;

to select a frame of the active speech data in dependence on characteristics of the active speech data reflected in the frame; and

to adjust the synchronization delay by adding or removing one or more audio samples in the selected frame so as not to significantly degrade the quality of the played back audio data.

10. Apparatus as in claim 9, where said control circuitry determines the number of audio samples in steps of size one sample.

11. Apparatus as in claim 9, where when the synchronization delay is adjusted by more than one audio sample, the adjustment is made by all of the determined audio samples in one adjustment.

12. Apparatus as in claim 9, where when the synchronization delay is adjusted by more than one audio sample, the adjustment is made by less than all of the determined audio samples by a plurality of adjustments.

13. Apparatus as in claim 9, where said control circuitry is further configured to select, if possible, an unvoiced frame.

14. Apparatus as in claim 9, where said control circuitry is further configured to discriminate against a transient frame for making the adjustment.

15. Apparatus as in claim 9, where said control circuitry, when determining by how much the synchronization delay should be adjusted, operates to measure an average amount of time that a frame resides in said buffer, and adjusts the synchronization delay so that the average duration approaches a desired buffer residency duration.

16. Apparatus as in claim 9, where at least one frame of audio data has a playback duration in the range of about 20 milliseconds to about 60 milliseconds.

17. Apparatus as in claim 9, where said buffer and control circuitry is contained within a wireless communications device, and where the packet is received from a radio channel.

18. Apparatus as in claim 9, where said buffer and control circuitry is contained within a device that processes and plays back packetized speech data.

19. Apparatus as in claim 9, where said buffer and control circuitry comprises part of a mobile telephone or a personal communicator.

20. Apparatus as in claim 9, where said buffer and control circuitry comprises part of a cellular radiotelephone.

21. Apparatus as in claim 9, where said buffer and control circuitry comprises part of a PC-based telephony system.

22. Apparatus as in claim 9, where said buffer and control circuitry comprises part of an IP telephony gateway.

23. Apparatus as in claim 9, where said buffer and control circuitry comprises part of an IP-to-circuit switched media transcoder.

24. A method for operating a communication device while synthesizing speech from speech data, the method operating to reduce synchronization delay and comprising:

for a received frame comprising encoded active speech data to be sent to a speech decoder, measuring the synchronization delay;

determining by how much the synchronization delay should be adjusted; and

adjusting the synchronization delay by adding or removing one or more speech samples in a selected frame of the active speech data so as not to significantly degrade the quality of the reproduced speech, where the frame is selected based on at least one speech decoder-related parameter so as to select, if possible, an unvoiced frame over a voiced frame, while discriminating against selecting a transient frame.

25. A method as in claim 24, where the step of determining determines the number of samples in steps of size one sample.

26. A method as in claim 24, where when the synchronization delay is adjusted by more than one sample, the adjustment is made by all of the determined samples in one adjustment.

27. A method as in claim 24, where when the synchronization delay is adjusted by more than one sample, the adjustment is made by less than all of the determined samples by a plurality of adjustments.

28. A method as in claim 24, where the step of determining comprises measuring an average amount of time that a frame resides in a jitter buffer;

and adjusting the synchronization delay so that the average duration approaches a desired jitter buffer residency duration.

29. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a pitch period.

30. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a pitch gain.

31. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a zero crossing rate within a received frame.

32. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of an energy distribution between adaptive and fixed codebook contributions.

33. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a measure of energy of a synthesized speech signal.

34. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a value of a linear prediction error.

35. A method as in claim 24, where the at least one speech decoder-related parameter is comprised of a value of a ratio of between an excitation signal at a synthesis filter input and the energy of a synthesized speech signal.

36. A method as in claim 24, where said speech decoder comprises a GSM speech decoder.