Audio signal processing method and encoder
An embodiment of the present invention discloses a data processing method, including: twiddling input data, so as to obtain twiddled data; pre-rotating the twiddled data by using a symmetric rotate factor, where the rotate factor is a·W 4L 2p+1 , p=0, . . . , L/2−1, and a is a constant; performing a Fast Fourier (Fast Fourier Transform, FFT) transform of L/2 point on the pre-rotated data, where L is the length of the input data; post-rotating the data that has undergone the FFT transform by using a symmetric rotate factor, where the rotate factor is b·W 4L 2q+1 , q=0, . . . , L/2−1, and b is a constant; and obtaining output data.
1. A method for processing an audio signal, the method comprising:
obtaining, by an encoder comprising a processor, a windowed time domain data by windowing an input audio signal;
obtaining, by the encoder, a twiddled signal based on the windowed time domain data;
pre-rotating, by the encoder, the twiddled signal by using a first symmetric rotation factor to obtain a pre-rotated data, wherein the first symmetric rotation factor is a·W 4L 2p+1 , p=0, . . . , L/2−1, wherein a is a constant, wherein L is the length of the input audio signal;
performing, by the encoder, a Fast Fourier transform (FFT) of L/2 points on the pre-rotated data to obtain FFT data;
performing, by the encoder, an in-place fixed rotate compensation on the FFT data;
post-rotating, by the encoder, the data that has undergone the in-place fixed rotate compensation by using a second symmetric rotation factor to obtain a post-rotated data, wherein the second symmetric rotation factor is b·W 4L 2p+1 , q=L/2−1, and b is a constant;
quantizing, by the encoder, a signal derived from the post-rotated data to obtain a quantized signal; and
writing, by the encoder, the quantized signal into a bitstream for transmitting or storing.
2. The method according to claim 1 , wherein the in-place fixed rotate compensation is performed by multiplying a fixed rotate compensation factor with the FFT data.
3. The method according to claim 2 , wherein the fixed rotate compensation factor is
1
+
j
(
3
π
4
L
)
.
4. The method according to claim 1 , wherein the twiddled signal is obtained according to z(p)={tilde over (x)}(2p)+j·{tilde over (x)}(L−1−2p), wherein z(p) denotes the twiddled signal, {tilde over (x)}(n) denotes the windowed time domain data.
5. The method according to claim 1 , wherein W 4L 2q+1 in the first symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
6. The method according to claim 1 , wherein W 4L 2q+1 in the second symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
7. An encoder for processing an audio signal, the encoder comprising:
a memory for storing processor-executable instructions; and
a processor operatively coupled to the memory, the processor being configured to execute the processor-executable instructions to cause the encoder to:
obtain a windowed time domain data by windowing an input audio signal,
obtain a twiddled signal based on the windowed time domain data;
pre-rotate the twiddled signal by using a first symmetric rotation factor to obtain a pre-rotated data, wherein the first symmetric rotation factor is a·W 4L 2p+1 , p=0, . . . , L/2−1, wherein a is a constant, wherein L is the length of the input audio signal,
perform a Fast Fourier transform (FFT) of L/2 points on the pre-rotated data to obtain FFT data,
perform an in-place fixed rotate compensation on the FFT data,
post-rotate the data that has undergone the in-place fixed rotate compensation by using a second symmetric rotate factor to obtain a post-rotated data, wherein the second symmetric rotation factor is b·W 4L 2q+1 , q=0, . . . , L/2−1, and b is a constant,
quantize a signal derived from the post-rotated data to obtain a quantized signal, and
write the quantized signal into a bitstream for transmitting or storing.
8. The encoder according to claim 7 , wherein the in-place fixed rotate compensation is performed by multiplying a fixed rotate compensation factor with the FFT data.
9. The method according to claim 8 , wherein the fixed rotate compensation factor is
1
+
j
(
3
π
4
L
)
.
10. The encoder according to claim 7 , wherein the twiddled signal is obtained according to z(p)={tilde over (x)}(2p)+j·{tilde over (x)}(L−1−2p), wherein z(p) denotes the twiddled signal, {tilde over (x)}(n) denotes the windowed time domain data.
11. The encoder according to claim 7 , wherein W 4L 2q+1 in the first symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
12. The encoder according to claim 7 , wherein W 4L 2q+1 in the second symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
13. A method for processing an audio signal, the method comprising:
obtaining, by an encoder comprising a processor, a windowed time domain data by windowing an input audio signal;
obtaining, by the encoder, a twiddled signal based on the windowed time domain data;
pre-rotating, by the encoder, the twiddled signal by using a first symmetric rotation factor to obtain a pre-rotated data, wherein the first symmetric rotation factor is a·W 4L 2p+1 , p=0, . . . , L/2−1, wherein a is a constant, wherein L is the length of the input audio signal;
performing, by the encoder, a Fast Fourier transform (FFT) of L/2 points on the pre-rotated data to obtain FFT data;
performing, by the encoder, an in-place fixed rotate compensation on the FFT data;
post-rotating, by the encoder, the data that has undergone the in-place fixed rotate compensation by using a second symmetric rotation factor to obtain a post-rotated data, wherein the second symmetric rotation factor is b·W 4L 2q+1 , q=0, . . . , L/2−1, and b is a constant;
obtaining, by the encoder, a frequency domain data based on the post-rotated data;
quantizing, by the encoder, a signal derived from the frequency domain data to obtain a quantized signal; and
writing, by the encoder, the quantized signal into a bitstream for transmitting or storing.
14. The method according to claim 13 , wherein the in-place fixed rotate compensation is performed by multiplying a fixed rotate compensation factor with the FFT data.
15. The method according to claim 14 , wherein the fixed rotate compensation factor is
1
+
j
(
3
π
4
L
)
.
16. The method according to claim 13 , wherein the twiddled signal is obtained according to z(p)={tilde over (x)}(2p)+j·{tilde over (x)}(L−1−2p), wherein z(p) denotes the twiddled signal, {tilde over (x)}(n) denotes the windowed time domain data.
17. The method according to claim 13 , wherein W 4L 2q+1 in the first symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
18. The method according to claim 13 , wherein W 4L 2q+1 in the second symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
19. The method according to claim 13 , wherein the frequency domain data is obtained by
{
y
(
2
q
)
=
Re
{
Z
_
(
q
)
}
,
y
(
L
-
1
-
2
q
)
=
-
Im
(
Z
_
(
q
)
}
,
q
=
0
,
…
,
L
/
2
-
1
;
wherein y(2q) is the real parts of the post-rotated data and denote the odd number frequency bins of the frequency domain data, y(L−1−2q) is the imaginary parts of the post-rotated data and denote the even number frequency bins of the frequency domain data, and
Z
_
(
q
)
=
(
1
+
j
3
π
/
4
L
)
*
W
4
L
2
q
+
1
∑
p
=
0
L
/
2
-
1
{
z
(
p
)
*
W
4
L
2
p
+
1
}
W
L
/
2
pq
.
20. An encoder for processing an audio signal, the encoder comprising:
a memory for storing processor-executable instructions; and
a processor operatively coupled to the memory, the processor being configured to execute the processor-executable instructions to cause the encoder to:
obtain a windowed time domain data by windowing an input audio signal,
obtain a twiddled signal based on the windowed time domain data,
pre-rotate the twiddled signal by using a first symmetric rotation factor to obtain a pre-rotated data, wherein the first symmetric rotation factor is a·W 4L 2q+1 , p=0, . . . , L/2−1, wherein a is a constant, wherein L is the length of the input audio signal,
perform a Fast Fourier transform (FFT) of L/2 points on the pre-rotated data to obtain FFT data,
perform an in-place fixed rotate compensation on the FFT data,
post-rotate the data that has undergone the in-place fixed rotate compensation by using a second symmetric rotation factor to obtain a post-rotated data, wherein the second symmetric rotation factor is b·W 4L 2q+1 , q=0, . . . , L/2−1, and b is a constant,
obtain a frequency domain data based on the post-rotated data,
quantize a signal derived from the frequency domain data to obtain a quantized signal, and
write the quantized signal into a bitstream for transmitting or storing.
21. The encoder according to claim 20 , wherein the in-place fixed rotate compensation is performed by multiplying a fixed rotate compensation factor with the FFT data.
22. The encoder according to claim 21 , wherein the fixed rotate compensation factor is
1
+
j
(
3
π
4
L
)
.
23. The encoder according to claim 21 , wherein the twiddled signal is obtained according to z(p)={tilde over (x)}(2p)+j·{tilde over (x)}(L−1−2p), wherein z(p) denotes the twiddled signal, {tilde over (x)}(n) denotes the windowed time domain data.
24. The encoder according to claim 21 , wherein W 4L 2q+1 in the first symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
25. The encoder according to claim 21 , wherein W 4L 2q+1 in the second symmetric rotation factor is expressed in the following form:
W
4
L
2
p
+
1
=
cos
2
π
(
2
p
+
1
)
4
L
-
j
sin
2
π
(
2
p
+
1
)
4
L
.
26. The encoder according to claim 21 , wherein the frequency domain data is obtained by
{
y
(
2
q
)
=
Re
{
Z
_
(
q
)
}
,
y
(
L
-
1
-
2
q
)
=
-
Im
(
Z
_
(
q
)
}
,
q
=
0
,
…
,
L
/
2
-
1
;
wherein y(2q) is the real parts of the post-rotated data and denote the odd number frequency bins of the frequency domain data, y(L−1−2q) is the imaginary parts of the post-rotated data and denote the even number frequency bins of the frequency domain data, and
Z
_
(
q
)
=
(
1
+
j
3
π
/
4
L
)
*
W
4
L
2
q
+
1
∑
p
=
0
L
/
2
-
1
{
z
(
p
)
*
W
4
L
2
p
+
1
}
W
L
/
2
pq
.