# US 2020/0243098 A1 · Audio Encoding/Decoding based on an Efficient Representation of Auto-Regressive Coefficients

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0243098 A1
- Document ID: US20200243098A1
- Status: Published application
- Kind: A1
- Application number: 16/832,597
- Filed: 2020-03-27
- Published: 2020-07-30
- Earliest claimed: 2011-11-02
- Assignees: Telefonaktiebolaget LM Ericsson (publ)
- Applicants: Telefonaktiebolaget LM Ericsson (publ)
- Inventors: Volodya Grancharov; Sigurdur Sverrisson
- Web page: https://patentgrep.com/patent/US20200243098A1

## CPC

- G10L 19/038: Vector quantisation, e.g. TwinVQ audio
- G10L 19/0204: {using subband decomposition}
- G10L 19/032: Quantisation or dequantisation of spectral components
- G10L 19/06: Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients
- G10L 21/038: using band spreading techniques
- G10L 2019/001: {Interpolation of codebook vectors}
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

An encoder for encoding a parametric spectral representation (ƒ) of auto-regressive coefficients that partially represent an audio signal. The encoder includes a low-frequency encoder configured to quantize elements of a part of the parametric spectral representation that correspond to a low-frequency part of the audio signal. It also includes a high-frequency encoder configured to encode a high-frequency part (ƒH) of the parametric spectral representation (ƒ) by weighted averaging based on the quantized elements ({circumflex over (ƒ)}L) flipped around a quantized mirroring frequency ({circumflex over (ƒ)}m) , which separates the low-frequency part from the high-frequency part, and a frequency grid determined from a frequency grid codebook in a closed-loop search procedure. Described are also a corresponding decoder, corresponding encoding/decoding methods and UEs including such an encoder/decoder.

## Claims

### Claim 1

A method, comprising:
encoding an audio signal, wherein encoding the audio signal comprises
obtaining a parametric spectral representation (ƒ) of auto-regressive coefficients (a) that partially represent the audio signal,
encoding a low-frequency part (ƒL) of the parametric spectral representation (ƒ) by quantizing coefficients of the parametric spectral representation that correspond to a low-frequency part of the audio signal, and
encoding a high-frequency part (ƒH) of the parametric spectral representation (ƒ) by weighted averaging based on the quantized coefficients ({circumflex over (ƒ)}L) flipped around a quantized mirroring frequency ({circumflex over (ƒ)}m), which separates the low-frequency part from the high-frequency part, and a frequency grid codebook obtained in a closed-loop search procedure; and
outputting, for transmission to a decoder, at least one quantitation index (IƒL) representing the quantized coefficients ({circumflex over (ƒ)}L), a quantization index (Im) representing the quantized mirroring f frequency ({circumflex over (ƒ)}m) and a quantization index (Ig) representing a frequency grid (gopt).

### Claim 2 (depends on 1)

The method of claim 1, further comprising transmitting encoded audio to a decoder, the encoded audio comprising the at least one quantitation index (IƒL), the quantization index (Im), and the quantization index (Ig).

### Claim 3 (depends on 1)

The method of claim 1, wherein encoding the audio signal further comprises quantizing the mirroring frequency {circumflex over (ƒ)}m , in accordance with:
{circumflex over (ƒ)}m=Q(ƒ(M/2)−{circumflex over (ƒ)}(M/2−1))+{circumflex over (ƒ)}(M/2−1),
where
Q denotes quantization of the expression in the adjacent parenthesis,
M denotes the total number of coefficients in the parametric spectral representation,
ƒ(M/2) denotes the first coefficient in the high-frequency part, and
{circumflex over (ƒ)}(M/2−1) denotes the last quantized coefficient in the low-frequency part.

### Claim 4 (depends on 3)

The method of claim 3, wherein encoding the audio signal further comprises flipping the quantized coefficients of the low frequency part (ƒL) of the parametric spectral representation (ƒ) around the quantized mirroring frequency {circumflex over (ƒ)}m in accordance with:
ƒflip(k)=2{circumflex over (ƒ)}m−{circumflex over (ƒ)}(M/2−1−k), 0≤k≤M/2−1,
where {circumflex over (ƒ)}(M/2−1−k) denotes quantized coefficient M/2−1−k.

### Claim 5 (depends on 4)

The method of claim 4, wherein encoding the audio signal further comprises rescaling the flipped coefficients ƒflip(k) in accordance with:
f
~
flip

(
k
)
=
{
(
f
flip

(
k
)
-
f
flip

(
0
)
)
·
(
f
max
-
f
^
m
)
/
f
^
m
+
f
flip

(
0
)

,
f
^
m
>
0
.
2

5
f
flip

(
k
)

,
otherwise
.

### Claim 6 (depends on 5)

The method of claim 5, wherein encoding the audio signal further comprises rescaling the frequency grids gi from the frequency grid codebook to fit into the interval between the last quantized coefficient {circumflex over (ƒ)}(M/2−1) in the low-frequency part and a maximum grid point value gmax in accordance with:
{tilde over (g)}i(k)=gi(k)·(gmax−{circumflex over (ƒ)}(M/2−1))+{circumflex over (ƒ)}(M/2−1).

### Claim 7 (depends on 6)

The method of claim 6, wherein encoding the audio signal further comprises weighted averaging of the flipped and rescaled coefficients {tilde over (ƒ)}flip(k) and the rescaled frequency grids {tilde over (g)}i(k) in accordance with:
ƒsmooth i(k)=[1−λ(k)]{tilde over (ƒ)}flip(k)+λ(k){tilde over (g)}i(k)
where λ(k) and [1−λ(k)] are predefined weights.

### Claim 8 (depends on 7)

The method of claim 7, wherein encoding the audio signal further comprises selecting a frequency grid gopt, where the index opt satisfies the criterion:
opt
=
arg

min
i

(
∑
k
=
0
M
/
2
-
1

(
f
smooth
i

(
k
)
-
f
H

(
k
)
)
2
)
where ƒH(k) is a target vector formed by the coefficients of the high-frequency part of the parametric spectral representation.

### Claim 9 (depends on 8)

The method of claim 8, wherein M=10, gmax=0.5, and the weights λ(k) are defined as λ={.0.2, 0.35, 0.5, 0.75, 0.8 }.

### Claim 10 (depends on 1)

The method of claim 1, wherein the encoding of the parametric spectral representation (ƒ) of auto-regressive coefficients is performed on a line spectral frequencies representation of the auto-regressive coefficients.

### Claim 11

An encoding apparatus, comprising:
an audio encoding circuit configured to:
encode an audio signal by
obtaining a parametric spectral representation (ƒ) of auto-regressive coefficients (a) that partially represent the audio signal,
encoding a low-frequency part (ƒL) of the parametric spectral representation (ƒ) by quantizing coefficients of the parametric spectral representation that correspond to a low-frequency part of the audio signal, and
encoding a high-frequency part (ƒH) of the parametric spectral representation (ƒ) by weighted averaging based on the quantized coefficients ({circumflex over (ƒ)}L) flipped around a quantized mirroring frequency ({circumflex over (ƒ)}m), which separates the low-frequency part from the high-frequency part, and a frequency grid codebook obtained in a closed-loop search procedure; and
output, for transmission to a decoder, at least one quantitation index (IƒL) representing the quantized coefficients ({circumflex over (ƒ)}L), a quantization index (Im) representing the quantized mirroring f frequency ({circumflex over (ƒ)}m), and a quantization index (Ig) representing a frequency grid (gopt).

### Claim 12 (depends on 11)

The encoding apparatus of claim 11, further comprising output circuitry configured to transmit encoded audio to a decoder, the encoded audio comprising the at least one quantitation index (IƒL) the quantization index (Im), and the quantization index (Ig).

### Claim 13 (depends on 11)

The encoding apparatus of claim 11, wherein the audio encoding circuit is further configured to quantize the mirroring frequency {circumflex over (ƒ)}m in accordance with:
{circumflex over (ƒ)}m=Q(ƒ(M/2)−{circumflex over (ƒ)}(M/2−1))+{circumflex over (ƒ)}(M/2−1),
where
Q denotes quantization of the expression in the adjacent parenthesis,
M denotes the total number of coefficients in the parametric spectral representation,
ƒ(M/2) denotes the first coefficient in the high-frequency part, and
{circumflex over (ƒ)}(M/2−1) denotes the last quantized coefficient in the low-frequency part.

### Claim 14 (depends on 13)

The encoding apparatus of claim 13, wherein the audio encoding circuit is further configured to flip the quantized coefficients of the low frequency part (ƒL)of the parametric spectral representation (ƒ) around the quantized minoring frequency {circumflex over (ƒ)}m, in accordance with:
ƒflip(k)=2{circumflex over (ƒ)}m−{circumflex over (ƒ)}(M/2−1−k), 0≤k≤M/2 −1.
where {circumflex over (ƒ)}(M/2−1−k) denotes the quantized coefficient M/2−1−k.

### Claim 15 (depends on 14)

The encoding apparatus of claim 14, wherein the audio encoding circuit is further configured to rescale the flipped coefficients ƒflip(k) in accordance with:
f
~
flip

(
k
)
=
{
(
f
flip

(
k
)
-
f
flip

(
0
)
)
·
(
f
max
-
f
^
m
)
/
f
^
m
+
f
flip

(
0
)

,
f
^
m
>
0
.
2

5
f
flip

(
k
)

,
otherwise

### Claim 16 (depends on 15)

The encoding apparatus of claim 15, wherein the audio encoding circuit is further configured to rescale the frequency grids gi from the frequency grid codebook to fit into the interval between the last quantized coefficient {circumflex over (ƒ)}(M/2/−1) in the low-frequency part and a maximum grid point value gmax in accordance with:
{tilde over (g)}i(k)=gi(k)·(gmax−{circumflex over (ƒ)}(M/2−1))+{circumflex over (ƒ)}(M/2−1).

### Claim 17 (depends on 16)

The encoding apparatus of claim 16, wherein the audio encoding circuit is further configured to perform weighted averaging of the flipped and rescaled coefficients {tilde over (ƒ)}flip(k)and the rescaled frequency grids {tilde over (g)}i(k) in accordance with:
ƒsmooth i(k)=[1−λ(k)]{tilde over (ƒ)}flip(k)+λ(k){tilde over (g)}i(k)
where λ(k) and [1−λ(k)] are prefedined weights.

### Claim 18 (depends on 17)

The encoding apparatus of claim 17, wherein the audio encoding circuit is further configured to select a frequency grid gopt, where the index opt satisfies the criterion:
opt
=
arg

min
i

(
∑
k
=
0
M
/
2
-
1

(
f
smooth
i

(
k
)
-
f
H

(
k
)
)
2
)
where ƒH(k) is a target vector formed by the coefficients of the high-frequency part of the parametric spectral representation.

### Claim 19 (depends on 18)

The encoding apparatus of claim 18, wherein M=10, gmax=0.5, and the weights λ(k) are defined as {=0.2, 0.35. 0.5, 0.75, 0.8}.

### Claim 20 (depends on 11)

The encoding apparatus of claim 11, wherein the audio encoding circuit is configured perform encoding of the parametric spectral representation (ƒ) of auto-regressive coefficients on a line spectral frequencies representation of the auto-regressive coefficients.
