Method and apparatus in a telecommunications system

• US 6,873,954 B1
• Filed: 09/05/2000
• Issued: 03/29/2005
• Est. Priority Date: 09/09/1999
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Sundqvist, Jim, Nohlgren, Anders, Frankkila, Tomas
• Primary Examiner(s): Chawan, Vijay B.
• Application Number:
  US09/655,326
• Time in Patent Office: 1,666 Days
• Field of Search: 704/258, 704/262, 704/264, 704/268, 704/219, 704/205, 704/207, 704/220
• US Class Current: 704/262
• 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

Abstract

Audio artifacts due to overrun or underrun in a playout buffer caused by the sampling rates at a sending and receiving side not being at the same rate are reduced. An LPC-residual is modified on a sample-by-sample basis. The LPC-residual block, which includes N samples, is converted to a block comprising N+1 or N−1 samples. A sample rate controller decides whether samples should be added to or removed from the LPC-residual. The exact position at which to add respective remove samples is either chosen arbitrarily or found by searching for low energy segments in the LPC-residual. A speech synthesiser module then reproduces the speech. By using the proposed sample rate conversion method the playout buffer can be continuously controlled. Furthermore, since the method works on a sample-by-sample basis the buffer can be kept to a minimum and hence no extra delay is introduced.

Claims

1. A method of improving speech quality in a communication system comprising a first terminal unit (TRX1), which transmits speech signals having a first sampling frequency (F1) and a second terminal (TRX2), which receives the speech signals, the method performed at the second terminal and comprising:

receiving said speech signals;

decoding the received speech frame;

buffering said decoded speech frame in a playout buffer of said second terminal (TRX2);

performing a dynamic sample rate conversion of said decoded speech frame comprising N samples on a sample by sample basis, said dynamic sample rate conversion comprising;

creating a first LPC-residual excitation frame comprising N samples derived from said decoded speech frame;

calculating whether a sample should be either added or removed from said first LPC-residual excitation frame;

selecting, in response to a determination that said calculating so demands, the position where in said first LPC-residual excitation frame to add or remove a sample;

generating a second modified LPC-residual excitation frame comprising at least one of N−

1 and N+1 samples, in response to a determination that said calculating so demands; and

synthesizing, in response to a determination that said calculating so demands, a second speech frame from said second modified LPC-residual excitation frame; and

playing out, in response to a determination that said calculating so demands, said second speech frame from said play out buffer.

2. The method of claim 1 wherein the creating step comprises performing an LPC-analysis of the speech frame in order to find LPC-parameters of said speech frame.

3. The method of claim 1 wherein the creating step comprises using already existing LPC-parameters from a speech decoder.

4. The method of claim 1 wherein the creating step comprises using an existing LPC-residual from a decoder.

5. The method of claim 1, wherein the calculating step comprises deciding whether a sample should be added or removed based on at least one of the following inputs:

sample frequencies of the sending (TRX1) and receiving (TRX2) terminal units;

a voice activity detector signal;

a status of the playout buffer, and

an indicator of a beginner of a talkspurt.

6. The method of claim 5 wherein said second second terminal unit uses a second sampling frequency (F2) which is a different frequency than said first sampling frequency.

7. The method of claim 1, wherein said position is found by searching for a segment of said first LPC-residual excitation frame with low energy.

8. The method of claim 7 wherein said segment of low energy is found via a block energy analysis.

9. The method of claim 7 wherein said segment of low energy is found via a sliding window energy analysis.

10. The method of claim 1 wherein said second second terminal unit uses a second sampling frequency (F2) which is a different frequency than said first sampling frequency.

11. The method of claim 1 wherein said position is found using knowledge about a position of a pitch pulse and knowledge about a time difference between said pitch pulse and a follow pitch pulse to select the position at which to add or remove a sample in the LPC-residual.

12. The method of claim 11 further comprising the step of finding said pitch pulse via a search for positions in the LPC residual with high energy.

13. The method of claim 12 wherein said positions with high energy are found via a block energy analysis.

14. The method of claim 12 wherein said positions with high energy are found via a sliding window energy analysis.

15. The method of claim 1 wherein said adding comprises adding a zero sample.

16. The method of claim 1 wherein said adding comprises adding a zero sample and interpolating surrounding samples.

17. The method of claim 1 wherein said removing comprises removing a sample from the LPC-residual.

18. The method of claim 1 wherein said adding comprises:

adding a sample in history of the LPC residual; and

increasing a lag pointer so long as the adding is within an LPC residual history.

19. The method of claim 1 wherein said removing comprises:

removing a sample in a history of the LPC residual; and

decreasing a lag pointer so long as the removing is within the LPC residual history.

20. The method of claim 1 wherein the second terminal unit comprises an adaptive and a fixed codebook;

and

wherein said adding comprises;

adding a sample in an output from the adaptive codebook;

extending an output from the fixed codebook; and

increasing a lag pointer so long as the adding is within the LPC residual history.

21. The method of claim 1 wherein the second terminal unit comprises an adaptive and a fixed codebook;

and

wherein said removing comprises;

removing a sample in an output from the adaptive codebook;

shortening an output from the fixed codebook; and

decreasing a lag pointer so long as the removing is within the LPC residual history.

22. The method of claim 1 wherein the second terminal unit comprises a fixed codebook;

and

wherein said adding or removing comprises adding or removing a sample in an output from the fixed codebook.

23. An apparatus for improving speech quality in a communication system comprising a first terminal unit (TRX1) which transmits speech signals having a first sampling frequency (F₁) and a second terminal unit (TRX2), which receives said speech signals, said apparatus comprising:

means for receiving said speech signals;

means for decoding the received speech frame;

means for buffering said decoded speech frame in a playout buffer of said second terminal (TRX2);

means for performing a dynamic sample rate conversion of said decoded speech frame comprising N samples on an sample by sample basis, wherein said means for performing said dynamic sample rate conversion comprises;

means for creating a first LPC-residual excitation frame comprising N samples derived from said speech time;

means for calculating whether a sample should be added or removed from first said LPC-residual excitation frame;

means for selecting, in response to a determination that said calculating so demands, the position where in said first LPC-residual excitation frame to add or remove a sample;

means for generating a second modified LPC-residual excitation frame comprising at least one of N−

1 and N+1 samples in response to a determination that said calculating so demands; and

means for synthesizing a second speech frame from said second modified LPC-residual excitation frame in response to a determination that said calculating so demands; and

means for playing out said second speech frame from said play out buffer in response to a determination that said calculating so demands.

24. The apparatus of claim 23 wherein the means for creating comprises means for performing an LPC-analysis of the speech frame to find LPC-parameters of said speech frame.

25. The apparatus of claim 23 wherein the means for creating comprises means for using existing LPC-parameters from a speech decoder.

26. The apparatus of claim 23 wherein said position is found by searching for a segment of said first LPC-residual excitation frame with low energy.

27. The apparatus of claim 23, wherein the means for calculating comprises means for deciding if a sample should be added or removed on the basis of a function of at least one of the following inputs;

sample frequencies of sending and receiving terminal units;

a voice activity detector signal;

a status of the playout buffer; and

an indicator of a beginning of a talkspurt.

28. The apparatus of claim 27 wherein said second second terminal unit uses a second sampling frequency (F2) which is a different frequency than said first sampling frequency.

29. The apparatus of claim 27 wherein the means for searching comprises means for performing a block energy analysis.

30. The apparatus of claim 27 wherein the means for searching comprises means for performing a sliding window energy analysis.

31. The apparatus of claim 23 wherein said second second terminal unit uses a second sampling frequency (F2) which is a different frequency then said first sampling frequency.

32. The apparatus of claim 23 wherein the means for selecting comprises means for using knowledge about a position of a pitch pulse together with knowledge about a time difference between said pitch pulse and a following pitch pulse to select the position at which to add or remove a sample in the LPC-residual.

33. The apparatus of claim 32 wherein the means for using knowledge about pitch pulses comprises means for finding the pitch pulses by searching for positions in the LPC residual with high energy.

34. The apparatus of claim 33 wherein the means for finding pitch pulses comprises means for performing a block energy analysis.

35. The apparatus of claim 33 wherein the means for finding pitch pulses comprises means for performing a sliding window energy analysis.

36. The apparatus of claim 23 wherein the means for performing adding or removing comprises means for adding a zero sample.

37. The apparatus of claim 23 wherein the means for performing adding or removing comprises means for removing a sample from the LPC-residual.

38. The apparatus of claim 23 wherein the means for performing adding or removing comprises means for adding a zero sample and interpolating surrounding samples.

39. The apparatus of claim 23 wherein the means for performing adding or removing comprises:

means for adding a sample in a history of the LPC residual; and

means for increasing a lag pointer so long as the adding is within the LPC residual history.

40. The apparatus of claim 23 wherein the means for performing adding or removing comprises:

means for removing a sample in a history of the LPC residual; and

means for decreasing a lag pointer so long as the removing is within the LPC residual history.

41. The apparatus of claim 23 wherein the second terminal unit comprises:

an adaptive and a fixed codebook;

means for adding a sample in an output from the adaptive codebook;

means for extending an output from the fixed codebook; and

means for increasing a lag pointer so long as the adding is within the LPC residual history.

42. The apparatus of claim 23 wherein the second terminal unit comprises:

an adaptive and a fixed codebook;

means for removing a sample in an output from the adaptive codebook;

means for removing a sample in an output from the fixed codebook; and

means for decreasing a lag pointer so long as the removing is within the LPC residual history.

43. The apparatus of claim 23 wherein the second terminal unit comprises:

a fixed codebook; and

means for adding or removing a sample in an output from the fixed codebook.