# US 2015/0106083 A1 · Audio signal loudness determination and modification in the frequency domain

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

- Number: US 2015/0106083 A1
- Document ID: US20150106083A1
- Status: Published application
- Kind: A1
- Application number: 14/517,875
- Filed: 2014-10-19
- Published: 2015-04-16
- Earliest claimed: 2008-12-24
- Assignees: Dolby Laboratories Licensing Corporation
- Applicants: Dolby Laboratories Licensing Corporation
- Inventors: Michael J. Smithers
- Web page: https://patentgrep.com/patent/US20150106083A1

## CPC

- H03G 9/005: {of digital or coded signals}
- G10L 19/0204: {using subband decomposition}
- G10L 19/022: Blocking, i.e. grouping of samples in time; Choice of analysis windows; Overlap factoring
- H03G 9/025: {frequency-dependent volume compression or expansion, e.g. multiple-band systems (H03G9/10, H03G9/18 take precedence)}
- H03G: CONTROL OF AMPLIFICATION
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Methods of, apparatuses for, and non-transitory computer readable media having instructions thereon that when executed cause carrying out methods of determining and modifying the perceived loudness of a frequency domain audio signal where the frequency resolution, and corresponding temporal coverage of the frequency domain information is not constant. The frequency (and thus temporal) resolution of the perceived loudness processing is maintained constant at the longest block size. One method includes a block combiner and a loudness modification interpolator.

## Claims

### Claim 1

A method of using a signal processing apparatus to determine the perceived loudness of an audio signal, the method comprising:
accepting into the signal processing apparatus blocks of frequency-domain audio data that results from applying a transform to corresponding blocks of time samples of an audio signal, the blocks including blocks of least two distinct block sizes, including a smallest block size and a largest block size, each respective accepted block's size being the smallest block size or a respective integer multiple of the smallest block size, with the largest block size-being the largest of all block sizes of the accepted blocks;
for each respective block size smaller than the largest block size, the signal processing apparatus combining a plurality of accepted blocks having the respective block size to form a respective formed largest-block-size block of frequency domain information; and
the signal processing apparatus determining or accepting one or more perceived loudness parameters of the accepted blocks or delayed versions thereof, each parameter comprising a respective parameter value for each of a set of critical frequency bands at the frequency resolution corresponding to the largest block size, the one or more perceived loudness parameters including the critical-band power spectrum and/or the specific loudness of the accepted blocks or of delayed versions thereof for the set of critical frequency bands.

### Claim 2 (depends on 1)

A method as recited in claim 1, further comprising:
using the one or more perceived loudness parameters at the frequency resolution corresponding to the largest block size, determining by the signal processing apparatus of at least one perceived loudness modification applicable to frequency domain information at the corresponding frequency resolution,
for each accepted block size of data other than the largest block size, interpolating the determined at least one loudness modification for application to accepted or delayed accepted blocks of data having said each block size, and
applying the determined at least one loudness modification or at least one interpolated loudness modification to the accepted blocks of frequency domain data or delayed version thereof.

### Claim 3 (depends on 2)

A method as recited in claim 2, wherein the combining for each respective block size smaller than the largest block size includes:
interleaving frequency domain data at the respective smaller block size, and/or
replicating values of frequency domain data at the respective smaller block size; and/or
interpolating between values of frequency domain data at the respective smaller block size, and/or
for a particular block size that has the property that the largest block size is a particular multiple greater than two of the particular block size, combining a subgroup of frequency domain data of consecutive blocks that occur towards the end of the plurality of blocks and that have overall energy above a pre-determined threshold using interleaving to form frequency domain data of an intermediate size block, and combining the frequency domain data of an intermediate size block with the plurality of blocks of frequency domain data that do not have energy over the predetermined threshold either replication or interleaving to create the largest-block-size frequency domain data.

### Claim 4 (depends on 2)

A method as recited in claim 2, further comprising delaying the blocks of frequency domain audio data, such that the at least one loudness modification or interpolated modification is applied to delayed versions of the blocks of frequency domain data.

### Claim 5 (depends on 2)

A method as recited in claim 2, wherein each determined loudness modifications includes a corresponding set of gains, wherein the interpolating determines interpolated gains, and wherein applying the determined at least one loudness modification or interpolated loudness modification includes applying the set of gains or set of interpolated gains.

### Claim 6 (depends on 2)

A method as recited in claim 2, wherein the one or more perceived loudness parameters include the excitation of the audio data, and the specific loudness of the audio data, and wherein carrying out the perceived loudness processing includes one or more of loudness leveling, gain control, noise compensation, and/or dynamic equalization.

### Claim 7 (depends on 2)

A method as recited in claim 2, wherein the transform is a lapped transform applied to overlapping blocks of time samples.

### Claim 8 (depends on 7)

A method as recited in claim 7, wherein the transform is a short time discrete Fourier transform, a short time discrete sine transform, or a modified discrete cosine transform.

### Claim 9 (depends on 2)

A method as recited in claim 2, wherein the data is from coded audio data coded by a perceptual coder that uses more than one block size.

### Claim 10 (depends on 9)

A method as recited in claim 9, wherein the data is from coded audio data coded by a perceptual coder uses a small block size of 256 time domain samples and a long block size of 512 time domain samples.

### Claim 11 (depends on 9)

A method as recited in claim 9, wherein the accepted frequency domain data is from coded audio data coded by a perceptual coder uses a small block size of 256 time domain samples and a long block size of 2048 time domain samples.

### Claim 12

A non-transitory computer readable medium configured with instructions that when executed by at least one processor carry out a method of perceived loudness processing of audio data that includes frequency domain data determined by a lapped transform at a plurality of block sizes, the method comprising:
accepting blocks of frequency domain audio data at the plurality of block sizes;
for accepted blocks at a particular short block size shorter than a longest block size of the plurality of block sizes, combining a plurality of blocks of frequency domain data at the particular short block size to form a block of frequency domain data at the longest block size; and
carrying out the perceived loudness processing of the accepted blocks at the longest block size,
wherein carrying out the perceived loudness processing includes determining one or more perceived loudness parameters.

### Claim 13 (depends on 12)

A non-transitory computer readable medium as recited in claim 12, wherein the loudness processing further includes:
determining at least one loudness modification to apply to the accepted blocks of frequency domain data or delayed version thereof, the determining the at least one loudness modification using the one or more determined perceived loudness parameters, the method further comprising, for the at least one loudness modification determined at the formed longest block size, interpolating the determined at least one loudness modification for application to the accepted blocks of data at the particular short block size used to form the determined at least one loudness modification; and
applying the determined at least one loudness modification or at least one interpolated loudness modification to the accepted blocks of frequency domain data or delayed version thereof.

### Claim 14 (depends on 12)

A non-transitory computer readable medium as recited in claim 12, wherein the transform is a lapped transform.

### Claim 15 (depends on 12)

A non-transitory computer readable medium as recited in claim 12, wherein the data is from coded audio data coded by a perceptual coder that uses more than one block size.

### Claim 16

An apparatus for perceived loudness processing of audio data comprising frequency domain data determined by a transform at a plurality of block sizes, including a largest block size and one or more other block sizes smaller than the largest block size, the apparatus comprising:
a processing engine comprising one or more processors; and
a storage subsystem coupled to the processing engine comprising one or more non-transitory media having stored therein instructions that when executed by at least one processor of the one or more processors, causes
accepting into the apparatus blocks of frequency-domain audio data that results from applying the transform on corresponding blocks of time samples of an audio signal, the blocks including blocks of least two distinct block sizes, including a smallest block size and a largest block size, each respective accepted block's size being the smallest block size or a respective integer multiple of the smallest block size, with the largest block size-being the largest of all block sizes of the accepted blocks;
for each respective block size smaller than the largest block size, combining a plurality of accepted blocks having the respective block size to form a respective formed largest-block-size block of frequency domain information; and
determining or accepting one or more perceived loudness parameters of the accepted blocks or delayed versions thereof, each parameter comprising a respective parameter value for each of a set of critical frequency bands at the frequency resolution corresponding to the largest block size, the one or more perceived loudness parameters including the critical-band power spectrum and/or the specific loudness of the accepted blocks or of delayed versions thereof for the set of critical frequency bands;
determining at least one perceived loudness modification applicable to frequency domain information at the frequency resolution corresponding to the largest block size, the determining using the one or more perceived loudness parameters at the frequency resolution corresponding to the largest block size;
for each accepted block size of data other than the largest block size, interpolating the determined at least one loudness modification for application to accepted or delayed accepted blocks of data having said each block size, and
applying the determined at least one loudness modification or at least one interpolated loudness modification to the accepted blocks of frequency domain data or delayed version thereof.
