US 12,717,841 B2Granted

Systems and methods for filtering large audio libraries using perceptive distribution binning

AssigneeMIIR Audio Technologies, Inc.

InventorsRoger Dumas, Jon Beck, Aaron Prust, Gary Katz, Paul J. Moe, Daniel J. Levitin

Life of this patent

Estimating the term…

Loading patent…

Abstract

System and methods for filtering and sorting libraries of audio data and quantitatively processing audio data to generate metrics for use in filtering and sorting methods based on distribution bins that are, for example, derived using human listening perception. Examples include calculating composite metrics based on values of objective audio metrics for individual audio files of a large audio library, each objective audio metrics generating a distribution of values. Examples include assigning three or more bins for each value of the objective audio metrics such that the bins represent perceptually distinct groups and, for each composite metric, combining the bins of each objective audio metric of the composite metric to generate corresponding distribution bins of the values of the composite metric for the large audio library. Examples include filtering the large audio library using corresponding distribution bins for a plurality of distinct composite metrics.

Claims16

Dependent claims

A method of generating perceptually filterable music libraries, the method comprising:

processing, using a processor, a plurality of digital audio files using a first objective audio processing metric to generate a first output for each of the plurality of digital audio files, the first output comprising a value of the first objective audio processing metric;
processing, using a processor, the plurality of digital audio files using a second objective audio processing metric to generate a second output for each of the plurality of digital audio files, the second output comprising a value of the second objective audio processing metric;
generating first normalized outputs by rescaling a range of the first outputs of the plurality of digital audio files;
generating second normalized outputs by rescaling a range of the second outputs of the plurality of digital audio files;
calculating a first plurality of distribution bins of the first normalized outputs based on a distribution of the first normalized outputs in the plurality of digital audio files;
calculating a second plurality of distribution bins of the second normalized outputs based on a distribution of the second normalized outputs in the plurality of digital audio files;
generating a first intermediate value for each of the plurality of digital audio files by separating mapping, for each of the first plurality of distribution bins, values of the first normalized outputs to a new value based on a relative size of a respective distribution bin of the first plurality of distribution bins;
generating a second intermediate value for each of the plurality of digital audio files by separating mapping, for each of the second plurality of distribution bins, values of the second normalized outputs to a new value based on a relative size of a respective distribution bin of the second plurality of distribution bins;
generating a composite metric value for each of the plurality of digital audio files by combining the first and second intermediate values of the plurality of digital audio files;
and
calculating a plurality of distribution bins of the composite metric values for the plurality of digital audio files based on a distribution of the composite metric value in the plurality of digital audio files;
wherein the first and second object audio processing algorithms are related to a same perceptual acoustic characteristic.

9 more elements

A method of generating perceptually filterable music libraries, the method comprising:

processing, using a processor, a plurality of digital audio files using a first objective audio processing metric to generate a first output for each of the plurality of digital audio files, the first output comprising a value of the first objective audio processing metric;
processing, using a processor, the plurality of digital audio files using a second objective audio processing metric to generate a second output for each of the plurality of digital audio files, the second output comprising a value of the second objective audio processing metric;
generating first normalized outputs by rescaling a range of the first outputs of the plurality of digital audio files;
generating second normalized outputs by rescaling a range of the second outputs of the plurality of digital audio files such that the first and second normalized ranges have a same range;
calculating a first plurality of distribution bins of the first normalized outputs based on a distribution of the first normalized outputs in the plurality of digital audio files and human perception of the first objective audio processing metric;
calculating a second plurality of distribution bins of the second normalized outputs based on a distribution of the second normalized outputs in the plurality of digital audio files and human perception of the second objective audio processing metric, the second plurality having a same number of bins as the first plurality;
generating a first intermediate value for each of the plurality of digital audio files by separating mapping, for each of the first plurality of distribution bins, values of the first normalized outputs to a new value based on a relative size of a respective distribution bin,
generating a second intermediate value for each of the plurality of digital audio files by separating mapping, for each of the second plurality of distribution bins, values of the second normalized outputs to a new value based on a relative size of a respective distribution bin,
generating a composite metric value for each of the plurality of digital audio files by combining the first and second intermediate values of the plurality of digital audio files;
and
calculating a plurality of distribution bins of the composite metric values for the plurality of digital audio files based on a distribution of the composite metric value in the plurality of digital audio files;
wherein the first and second object audio processing algorithms are related to a same perceptual acoustic characteristic.

9 more elements

Dependencies are inferred from the claim text.