JITTER BUFFER AND JITTER BUFFER CONTROLLING METHOD

• US 20090316689A1
• Filed: 12/30/2008
• Published: 12/24/2009
• Est. Priority Date: 06/18/2008
• Status: Active Grant

• Current Assignee: Hon Hai Precision Industry Co., Ltd.
• Original Assignee: Hon Hai Precision Industry Co., Ltd.
• Inventors: YANG, JENG-SHYAN
• Granted Patent: US 8,243,721 B2
• Time in Patent Office: Days
• US Class Current: 370/352
• CPC Class Codes: H04L 65/80 Responding to QoS

Abstract

A jitter buffer controlling method includes a data writing step, a data buffering step and a data reading step. The data writing step and the data reading step are executed synchronously and repeatedly. The data writing step includes detecting whether a data packet that comprises a series of voice data frames is normally received, and calculating a storage address for each of the voice data frames. The data buffering step includes buffering the voice data frames, and storing each of the voice data frames in a corresponding storage address calculated in the data writing step. The data reading step includes transmitting the voice data frames to a voice digital signal processor (VDSP) for playing.

Claims

1. A method for controlling a jitter buffer, the method comprising:

a data writing step comprising;

receiving a data packet that comprises a series of voice data frames, detecting whether the data packet is normally received according to a packet timestamp of the data packet, a voice timestamp of a voice data frame that is just read out from the jitter buffer, and a delay depth of the jitter buffer, and calculating a storage address for each of the voice data frames contained in the data packet according to the detection result;

a data buffering step comprising;

buffering the voice data frames, and storing each of the voice data frames in a corresponding storage address calculated in the data writing step; and

a data reading step comprising;

transmitting each of the voice data frames to a voice digital signal processor (VDSP) for playing.

2. The method of claim 1, wherein the data writing step and the data reading step are executed synchronously and repeatedly.

3. The method of claim 1, wherein the data writing step further comprises:

detecting whether the data packet is the first received data packet;

saving the data packet into a pre-buffering phase of the jitter buffer, and setting a reference timestamp for the first received data packet, if the data packet is the first received data packet;

determining whether a difference between the packet timestamp and the reference timestamp is less than the delay depth of the jitter buffer, if the data packet is not the first received data packet;

determining that the data packet is in the pre-buffering phase and buffering the data packet, if the difference is less than the delay depth;

determining whether the packet timestamp is less than the voice timestamp, if the difference is not less than the delay depth;

discarding the data packet and increasing a tuning delay depth of the jitter buffer by adding an adjustment value, if the packet timestamp is less than the voice timestamp; and

determining whether the packet timestamp is greater than a sum value of the delay depth and the voice timestamp, if the packet timestamp is not less than the voice timestamp.

4. The method of claim 3, wherein the buffering step further comprises:

storing the voice data frames and the voice timestamp of each of the voice data frames in a first storage address, if the packet timestamp is not greater than the sum value; and

storing the voice data frames and the voice timestamp of each of the voice data frames in a second storage address if the packet timestamp is greater than the sum value, and increasing the tuning delay depth of jitter buffer with the adjustment value.

5. The method of claim 4, wherein the first storage address is calculated by a formula (TSw−

TSr)/SPF+ReadIdx, wherein TSw represents the packet timestamp, TSr represents the voice timestamp of the voice data frame that is just read out from the jitter buffer, SPF represents a time interval between the voice data frames are transmitted, and ReadIdx represents a storage address of the voice data frame that is at a read position of the jitter buffer.

6. The method of claim 4, wherein the second storage address is calculated by a formula Dn/SPF+ReadIdx, wherein Dn represents the delay depth, SPF represents a time interval between the voice data frames are transmitted, and ReadIdx represents a storage address of the voice data frame that is at a read position of the jitter buffer.

7. The method of claim 4, wherein the adjustment value equals SPF/2ⁿ, wherein SPF represents a time interval between the voice data frames are transmitted, and n is an integer that is between zero and eight.

8. The method of claim 3, further comprising:

determining whether an absolute value of the tuning delay depth is greater than a time interval between the voice data frames are transmitted; and

adjusting the delay depth of jitter buffer by increasing or decreasing the delay depth with one time interval if the absolute value of the tuning delay depth is greater than the time interval.

9. The method of claim 1, wherein the data reading step further comprises:

detecting whether the jitter buffer is in a pre-buffering phase when the VDSP starts to play one of the voice data frames;

sending a silence voice data frame to the VDSP if the jitter buffer is in the pre-buffering phase, or detecting whether the voice timestamp is null if the jitter buffer is not in the pre-buffering phase;

setting the voice timestamp of the voice data frame to a sum value of a former timestamp and a time interval between the voice data frames are transmitted, recording one voice data frame to be played, and executing a voice data frame compensation on the jitter buffer if the voice timestamp is null; and

sending the voice data frame that is at a read position of the jitter buffer to the VDSP and increasing a counter of the jitter buffer by one if the voice timestamp is not null.

10. The method of claim 9, wherein the voice data frame compensation comprises:

detecting whether the former timestamp of a former played voice data frame is null;

sending the former played voice data frame to the VDSP if the former timestamp is not null, or detecting whether a later timestamp of a later voice data frame is null if the former timestamp is null;

sending the later voice data frame to the VDSP if the later timestamp is not null, or detecting whether a former two timestamp of a former two voice data frame is null if the later timestamp is null; and

sending the former two voice data frame to the VDSP if the former two timestamp is not null, or sending a silence voice data frame to the VDSP if the former two timestamp is null.

11. A jitter buffer, comprising:

a data writing module configured for receiving a data packet that comprises a series of voice data frames, detecting whether the data packet is normally received according to a packet timestamp of the data packet, a voice timestamp of a voice data frame just read out from the jitter buffer, and a delay depth of the jitter buffer, and calculating a storage address for each of the voice data frames contained in the data packet according to the detection result;

a data buffering module configured for buffering the voice data frames, and storing each of the voice data frames in a corresponding storage address calculated by the data writing module; and

a data reading module configured for transmitting each of the voice data frames to a voice digital signal processor (VDSP) for playing.

12. The jitter buffer of claim 11, wherein the data writing module and the data reading module are executed synchronously and repeatedly.

13. The jitter buffer of claim 11, wherein the data reading module is further configured for detecting whether the jitter buffer is in a pre-buffering phase when the VDSP starts to play one of the voice data frames, and sending a silence voice data frame to the VDSP if the jitter buffer is in the pre-buffering phase.

14. The jitter buffer of claim 11, wherein the data reading module is further configured for setting the voice timestamp of the voice data frame to a sum value of a former timestamp and a time interval between the voice data frames are transmitted, recording one voice data frame to be played, and executing a voice data frame compensation on the jitter buffer if the voice timestamp is null.

15. The jitter buffer of claim 14, wherein the voice data frame compensation comprises:

detecting whether the former timestamp of a former played voice data frame is null;

sending the former played voice data frame to the VDSP if the former timestamp is not null, or detecting whether a later timestamp of a later voice data frame is null if the former timestamp is null;

sending the later voice data frame to the VDSP if the later timestamp is not null, or detecting whether a former two timestamp of a former two voice data frame is null if the later timestamp is null; and

sending the former two voice data frame to the VDSP if the former two timestamp is not null, or sending a silence voice data frame to the VDSP if the former two timestamp is null.