Acoustic fingerprinting

A non-transitory computer-readable medium storing computer-executable instructions that when executed by at least a processor of a computer cause the computer to:

• generate a target acoustic fingerprint from target acoustic output of a target device, wherein the target acoustic fingerprint includes a set of component time series signals sampled from a set of frequencies that have highest power spectral density peaks among frequencies of reference acoustic output of a reference device;
• compensate for ambient noise in the component time series signals based on values for the component time series signals predicted by a machine learning algorithm;
• generate a similarity metric that quantifies similarity of the target acoustic fingerprint to a reference acoustic fingerprint of the reference device, wherein the similarity metric is a cumulative mean absolute error between the component time series signals of the target acoustic fingerprint and corresponding component time series signals of the reference acoustic fingerprint;
• comparing the similarity metric to a threshold;
• in response to a first comparison result of the comparing the similarity metric to the threshold, indicating that the target device matches the reference device;
• and
• in response to a second comparison result of the comparing the similarity metric to the threshold, indicating that the target device does not match the reference device.

A computing system, comprising:

• a processor;
• and
• a memory operably connected to the processor;
• a non-transitory computer-readable medium operably connected to the processor and memory and storing computer-executable instructions that when executed by at least the processor cause the computing system to:
• generate a target acoustic fingerprint from measurements of acoustic output of a target device, wherein the target acoustic fingerprint includes a plurality of component time series signals sampled from a set of frequencies that have highest power spectral density peaks among frequencies of reference acoustic output of a reference device;
• compensate for ambient noise in the target acoustic fingerprint by substituting values of the component time series signals with machine learning predictions of the values;
• generate a similarity metric that quantifies similarity of the target acoustic fingerprint to a reference acoustic fingerprint of a reference device, wherein the similarity metric is a cumulative mean absolute error between the component time series signals of the target acoustic fingerprint and corresponding component time series signals of the reference acoustic fingerprint;
• compare the similarity metric to a threshold;
• in response to a first comparison result of the comparing the similarity metric to the threshold, indicating that the target device matches the reference device;
• and
• in response to a second comparison result of the comparing the similarity metric to the threshold, indicating that the target device does not match the reference device.