Audio coding method and related apparatus

• US 10,056,089 B2
• Filed: 01/18/2017
• Issued: 08/21/2018
• Est. Priority Date: 07/28/2014
• Status: Active Grant

• Current Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Inventors: Liu, Zexin, Miao, Lei
• Primary Examiner(s): Saint Cyr, Leonard
• Application Number:
  US15/408,442
• Publication Number:
  US 20170125031A1
• Time in Patent Office: 580 Days
• Field of Search: 704500-504
• CPC Class Codes: G10L 19/0204 using subband decomposition
• CPC Class Codes: G10L 19/0208 Subband vocoders
• CPC Class Codes: G10L 19/0212 using orthogonal transformation
• CPC Class Codes: G10L 19/12 the excitation function being a code excitation, e.g. in code excited linear prediction [CELP] vocoders
• CPC Class Codes: G10L 19/22 Mode decision, i.e. based on audio signal content versus external parameters
• CPC Class Codes: G10L 25/18 the extracted parameters being spectral information of each sub-band
• CPC Class Codes: G10L 25/21 the extracted parameters being power information

Abstract

An audio encoding method and a related apparatus are disclosed. The audio coding method includes: performing a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current audio frame; obtaining one or more reference coding parameters of the current frame; and determining whether the reference coding parameters satisfy a set of parameter conditions. If any one of the parameter conditions is satisfied, the spectral coefficients of the current frame are encoded by using a transform coded excitation (TCX) algorithm. If none of the parameter conditions is satisfied, the spectral coefficients of the current audio frame are encoded using a high quality transform coding (HQ) algorithm. The audio encoding method and the related apparatus help improve encoding quality or encoding efficiency in audio signal encoding.

Claims

1. A method for encoding an audio signal, performed by an audio signal encoder which comprises a processor and a memory storing program instructions for execution by the processor, the method comprising:

performing, by the encoder, a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtaining, by the encoder, one or more reference coding parameters of the current frame;

determining, by the encoder, whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein the current frame comprises a subband z, and two subbands i and j;

wherein the subband z and the subbands i and j are such that;

a highest frequency bin of the subband z is higher than a critical frequency bin F1, and F1 is in a range of 6.4 kHz to 12 kHz; and

a highest frequency bin of the subband i is lower than a highest frequency bin of the subband j, the highest frequency bin of the subband j is higher than a critical frequency bin F2, and F2 is in a range of 4.8 kHz to 8 kHz;

wherein the obtained reference coding parameters comprise;

a peak value of spectral coefficients that are located within the subband z, and an average value of spectral coefficients that are located within the subband z; and

an average energy of spectral coefficients that are located within the subband and an average energy of spectral coefficients that are located within the subband j;

and wherein the parameter conditions comprise a parameter condition as following;

the peak value of the spectral coefficients that are located within the subband z is greater than the average value of the spectral coefficients that are located within the subband z multiplied by a threshold T2; and

the average energy of the spectral coefficients that are located within the subband j is greater than a product of the average energy of the spectral coefficients that are located within the subband i multiplied by a threshold T4.

2. The method according to claim 1, wherein a lowest frequency bin of the subband z is higher than or equal to the critical frequency bin F1, and wherein

the highest frequency bin of the subband i is lower than or equal to a lowest frequency bin of the subband j;

or

a lowest frequency bin of the subband j is higher than the critical frequency bin F2.

3. The method according to claim 1, wherein the threshold T2 is no less than 1, or the threshold T2 is no less than 2, or the threshold T2 is no less than 3, or the threshold T2 is no less than 5;

and

wherein the threshold T4 is no less than 0.5, or the threshold T4 is no less than 1, or the threshold T4 is no less than 2, or the threshold T4 is no less than 3.

4. The method according to claim 1, wherein a range of frequency bins of the subband z is 8 kHz to 12 kHz, or 9 kHz to 11 kHz, or 8 kHz to 9.6 kHz, or 12 kHz to 14 kHz;

wherein a range of frequency bins of the subband i is 3.2 kHz to 6.4 kHz, or 3.2 kHz to 4.8 kHz, or 4.8 kHz to 6.4 kHz, or 0.4 kHz to 6.4 kHz, or 0.4 kHz to 3.6 kHz; and

wherein a range of frequency bins of the subband j is 6.4 kHz to 9.6 kHz, or 6.4 kHz to 8 kHz, or 8 kHz to 9.6 kHz, or 4.8 kHz to 9.6 kHz, or 4.8 kHz to 8 kHz.

5. A method for encoding an audio signal, performed by an audio signal encoder which comprises a processor and a memory storing program instructions for execution by the processor, the method comprising:

performing, by the encoder, a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtaining, by the encoder, one or more reference coding parameters of the current frame;

determining, by the encoder, whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein the current frame comprises subband i and subband j;

wherein a highest frequency bin of the subband i is lower than a highest frequency bin of the subband j, the highest frequency bin of the subband j is higher than a critical frequency bin F2, and F2 is in a range of 4.8 kHz to 8 kHz;

wherein the obtained reference coding parameters comprise;

an average energy of spectral coefficients that are located within the subband and an average energy of spectral coefficients that are located within the subband j;

and wherein the parameter conditions comprise a parameter condition as following;

the average energy of the spectral coefficients that are located within the subband j is greater than a product of the average energy of the spectral coefficients that are located within the subband i multiplied by a threshold T4.

6. The method according to claim 5, wherein the highest frequency bin of the subband i is lower than or equal to a lowest frequency bin of the subband j;

or a lowest frequency bin of the subband j is higher than the critical frequency bin F2.

7. The method according to claim 5, wherein the threshold T4 is no less than 0.5, or the threshold T4 is no less than 1, or the threshold T4 is no less than 2, or the threshold T4 is no less than 3.

8. The method according to claim 5, wherein a range of frequency bins of the subband i is 3.2 kHz to 6.4 kHz, or 3.2 kHz to 4.8 kHz, or 4.8 kHz to 6.4 kHz, or 0.4 kHz to 6.4 kHz, or 0.4 kHz to 3.6 kHz;

and

wherein a range of frequency bins of the subband j is 6.4 kHz to 9.6 kHz, or 6.4 kHz to 8 kHz, or 8 kHz to 9.6 kHz, or 4.8 kHz to 9.6 kHz, or 4.8 kHz to 8 kHz.

9. A method for encoding an audio signal, performed by an audio signal encoder which comprises a processor and a memory storing program instructions for execution by the processor, the method comprising:

performing, by the encoder, a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtaining, by the encoder, one or more reference coding parameters of the current frame;

determining, by the encoder, whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encoding the audio signal comprising;

quantizing, by the encoder, the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

writing the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein the current frame comprise subband x and subband y;

wherein a highest frequency bin of the subband x is lower than or equal to a lowest frequency bin of the subband y;

wherein the obtained reference coding parameters comprise;

a peak value of spectral coefficients that are located within the subband x, an average value of spectral coefficients that are located within the subband x, a peak value of spectral coefficients that are located within the subband y, and an average value of spectral coefficients that are located within the subband y;

and wherein the parameter conditions comprise a parameter condition as following;

a product of the peak value of spectral coefficients that are located within the subband x multiplied by the average value of coefficients that are located within the subband y is less than a product of the peak value of spectral coefficients that are located within the subband y multiplied by the average value of coefficients that are located within the subband x and multiplied by a lowest value of an interval R1;

or

a product of the peak value of spectral coefficients that are located within the subband x multiplied by the average value of coefficients that are located within the subband y is greater than a product of the peak value of spectral coefficients that are located within the subband y multiplied by the average value of coefficients that are located within the subband x and multiplied by a highest value of the interval R1.

10. The method according to claim 9, wherein the interval R1 is [0.5, 2], or the interval R1 is [0.4, 2.5], or the interval R1 is [0.8, 1.25].

11. The method according to claim 9, wherein a range of frequency bins of the subband x is 1 kHz to 2.6 kHz, and a range of frequency bins of the subband y is 4.8 kHz to 6.4 kHz.

12. An audio signal encoder, comprising a processor and a memory storing program instructions for execution by the processor;

wherein the processor is configured to execute the program instructions to;

perform a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtain one or more reference coding parameters of the current frame;

determine whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein the current frame comprises a subband z, and two subbands i and j;

wherein the subband z and the subbands i and j are such that

a highest frequency bin of the subband z is higher than a critical frequency bin F1, and F1 is in a range of 6.4 kHz to 12 kHz; and

a highest frequency bin of the subband i is lower than a highest frequency bin of the subband j, the highest frequency bin of the subband j is higher than a critical frequency bin F2, and F2 is in a range of 4.8 kHz to 8 kHz;

wherein the obtained reference coding parameters comprise;

a peak value of spectral coefficients that are located within the subband z, and an average value of spectral coefficients that are located within the subband z; and

an average energy of spectral coefficients that are located within the subband i, and an average energy of spectral coefficients that are located within the subband j;

and wherein the parameter conditions comprise a parameter condition as following;

the peak value of the spectral coefficients that are located within the subband z is greater than the average value of the spectral coefficients that are located within the subband z multiplied by a threshold T2; and

the average energy of the spectral coefficients that are located within the subband j is greater than a product of the average energy of the spectral coefficients that are located within the subband i multiplied by a threshold T4.

13. The audio signal encoder according to claim 12, wherein a lowest frequency bin of the subband z is higher than or equal to the critical frequency bin F1, and wherein

the highest frequency bin of the subband i is lower than or equal to a lowest frequency bin of the subband j;

or

a lowest frequency bin of the subband j is higher than the critical frequency bin F2.

14. The audio signal encoder according to claim 12, wherein the threshold T2 is no less than 1, or the threshold T2 is no less than 2, or the threshold T2 is no less than 3, or the threshold T2 is no less than 5;

and

wherein the threshold T4 is no less than 0.5, or the threshold T4 is no less than 1, or the threshold T4 is no less than 2, or the threshold T4 is no less than 3.

15. The audio signal encoder according to claim 12, wherein a range of frequency bins of the subband z is 8 kHz to 12 kHz, or 9 kHz to 11 kHz, or 8 kHz to 9.6 kHz, or 12 kHz to 14 kHz;

wherein a range of frequency bins of the subband i is 3.2 kHz to 6.4 kHz, or 3.2 kHz to 4.8 kHz, or 4.8 kHz to 6.4 kHz, or 0.4 kHz to 6.4 kHz, or 0.4 kHz to 3.6 kHz; and

wherein a range of frequency bins of the subband j is 6.4 kHz to 9.6 kHz, or 6.4 kHz to 8 kHz, or 8 kHz to 9.6 kHz, or 4.8 kHz to 9.6 kHz, or 4.8 kHz to 8 kHz.

16. An audio signal encoder, comprising a processor and a memory storing program instructions for execution by the processor;

wherein the processor is configured to execute the program instructions to;

perform a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtain one or more reference coding parameters of the current frame;

determine whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein the current frame comprises subband i and subband j;

wherein a highest frequency bin of the subband i is lower than a highest frequency bin of the subband j, the highest frequency bin of the subband j is higher than a critical frequency bin F2, and F2 is in a range of 4.8 kHz to 8 kHz;

wherein the obtained reference coding parameters comprise;

an average energy of spectral coefficients that are located within the subband and an average energy of spectral coefficients that are located within the subband j;

and wherein the parameter conditions comprise a parameter condition as following;

the average energy of the spectral coefficients that are located within the subband j is greater than a product of the average energy of the spectral coefficients that are located within the subband i multiplied by a threshold T4.

17. The audio signal encoder according to claim 16, wherein the highest frequency bin of the subband i is lower than or equal to a lowest frequency bin of the subband j;

or a lowest frequency bin of the subband j is higher than the critical frequency bin F2.

18. The audio signal encoder according to claim 16, wherein the threshold T4 is no less than 0.5, or the threshold T4 is no less than 1, or the threshold T4 is no less than 2, or the threshold T4 is no less than 3.

19. The audio signal encoder according to claim 16, wherein a range of frequency bins of the subband i is 3.2 kHz to 6.4 kHz, or 3.2 kHz to 4.8 kHz, or 4.8 kHz to 6.4 kHz, or 0.4 kHz to 6.4 kHz, or 0.4 kHz to 3.6 kHz;

and

wherein a range of frequency bins of the subband j is 6.4 kHz to 9.6 kHz, or 6.4 kHz to 8 kHz, or 8 kHz to 9.6 kHz, or 4.8 kHz to 9.6 kHz, or 4.8 kHz to 8 kHz.

20. An audio signal encoder, comprising a processor and a memory storing program instructions for execution by the processor;

wherein the processor is configured to execute the program instructions to;

perform a time-frequency transformation on a current frame of a time-domain audio signal, to obtain spectral coefficients of the current frame;

obtain one or more reference coding parameters of the current frame;

determine whether the reference coding parameters satisfy any one of a set of parameter conditions;

(1) when any one of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a transform coded excitation (TCX) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

or

(2) when none of the parameter conditions is satisfied, encode the audio signal comprising;

quantize the spectral coefficients of the current frame using a high quality transform coding (HQ) algorithm; and

write the quantized spectral coefficients into a bitstream for storing or transmitting;

wherein a highest frequency bin of the subband x is lower than or equal to a lowest frequency bin of the subband y;

wherein the obtained reference coding parameters comprise;

a peak value of spectral coefficients that are located within the subband x, an average value of spectral coefficients that are located within the subband x, a peak value of spectral coefficients that are located within the subband y, and an average value of spectral coefficients that are located within the subband y;

and wherein the parameter conditions comprise a parameter condition as following;

a product of the peak value of spectral coefficients that are located within the subband x multiplied by the average value of coefficients that are located within the subband y is less than a product of the peak value of spectral coefficients that are located within the subband y multiplied by the average value of coefficients that are located within the subband x and multiplied by a lowest value of an interval R1;

or

a product of the peak value of spectral coefficients that are located within the subband x multiplied by the average value of coefficients that are located within the subband y is greater than a product of the peak value of spectral coefficients that are located within the subband y multiplied by the average value of coefficients that are located within the subband x and multiplied by a highest value of the interval R1.

21. The audio signal encoder according to claim 20, wherein the interval R1 is [0.5, 2], or the interval R1 is [0.4, 2.5], or the interval R1 is [0.8, 1.25].

22. The audio signal encoder according to claim 20, wherein a range of frequency bins of the subband x is 1 kHz to 2.6 kHz, and a range of frequency bins of the subband y is 4.8 kHz to 6.4 kHz.