US 9,069,849 B1Granted

Methods for enforcing time alignment for speed resistant audio matching

AssigneeGoogle Inc.

InventorsDominik Roblek, Gheorghe Postelnicu, Matthew Sharifi

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods are provided herein relating to speed resistant audio matching. Descriptors can be generated for a received audio signal and matched with reference descriptors. A set of hits for respective reference samples can be generated based on the matching. A histogram can then be generated that correlates probe sample hit time with reference sample hit time. In one implementation, a rolling window can be used in analyzing the histogram allowing for slight variances in the timing between probe sample hits and reference sample hits. In another implementation, the histogram generated can be based on an estimated time stretch of the probe sample. In yet another implementation, a set of histograms can be generated based on a minimum speed change, a maximum speed change, and a speed step. Histograms can be evaluated to determine a most likely matching histogram.

Claims19

Dependent claims

A system comprising:

a memory that has stored thereon computer executable components;
and
a processor that executes the following computer executable components stored in the memory:
an input component that receives an audio sample;
a descriptor component that generates a set of descriptors based on the audio sample;
a matching component that generates respective sets of hits for reference samples among a set of reference samples based on comparing the set of descriptors to a set of reference descriptors associated with reference samples among the set of reference samples, wherein a hit indicates a match between an audio sample descriptor and a reference descriptor at a sample hit time and a reference hit time;
a histogram component that generates respective histograms for the reference samples based on the sample hit times and the reference hit times for hits in the respective sets of hits for the reference samples;
a rolling window component that rolls a window over the respective histograms for the reference samples and, upon completion of the rolling, determines a maximum number of windowed hits within the window at any point during the roll, wherein the width of the window is based on time stretch sensitivity, and wherein the rolling window component further extracts a subset of the set of hits, where the subset of the set of hits are hits within the window at the maximum number of windowed hits;
and
an evaluation component that evaluates sets of histograms to determine a most likely matching histogram.

7 more elements

A method comprising:

receiving, by a system including a processor, an audio sample;
generating, by the system, a set of descriptors based on the audio sample;
generating, by the system, respective sets of hits for reference samples among a set of reference samples based on comparing the set of descriptors to a set of reference descriptors associated with reference samples among the set of reference samples, wherein a hit indicates a match between an audio sample descriptor and a reference descriptor at a sample hit time and a reference hit time;
generating, by the system, respective histograms for the reference samples based on the sample hit time and the reference hit time for hits in the respective sets of hits for the reference samples;
rolling, by the system, a window over the respective histograms for respective reference samples wherein the width of the window is based on a time stretch sensitivity;
upon completion of the rolling, determining, by the system, a maximum number of windowed hits within the window at any point during the roll;
extracting, by the system, a subset of the set of hits wherein the subset of the set of hits are the hits within the window at the maximum number of windowed hits;
and
evaluating, by the system, sets of histograms to determine a most likely matching histogram.

6 more elements

A non-transitory computer-readable medium having instructions stored thereon that, in response to execution, cause a system including a processor to perform operations, comprising:

receiving an audio sample;
generating a set of descriptors based on the audio sample;
generating respective sets of hits for reference samples among a set of reference samples based on comparing the set of descriptors to a set of reference descriptors associated with reference samples among the set of reference samples, wherein a hit indicates a match between an audio sample descriptor and a reference descriptor at a sample hit time and a reference hit time;
generating respective histograms for the reference samples based on the sample hit time and the reference hit time for hits in the respective sets of hits for the reference samples;
rolling a window over the respective histograms for respective reference samples wherein the width of the window is based on a time stretch sensitivity;
upon completion of the rolling, determining a maximum number of windowed hits within the window at any point during the roll;
extracting a subset of the set of hits wherein the subset of the set of hits are the hits within the window at the maximum number of windowed hits;
and
evaluating sets of histograms to determine a most likely matching histogram.

6 more elements

Dependencies are inferred from the claim text.