US 9,728,188 B1Granted

Methods and devices for ignoring similar audio being received by a system

AssigneeAmazon Technologies, Inc.

InventorsAlexander David Rosen, Michael James Rodehorst, George Jay Tucker, Aaron Lee Mathers Challenner

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods for detecting similar audio being received by separate voice activated electronic devices, and ignoring those commands, is described herein. In some embodiments, a voice activated electronic device may be activated by a wakeword that is output by the additional electronic device, such as a television or radio, may capture audio of sound subsequently following the wakeword, and may send audio data representing the sound to a backend system. Upon receipt, the backend system may, in parallel to performing automated speech recognition processing to the audio data, generate a sound profile of the audio data, and may compare that sound profile to sound profiles of recently received audio data and/or flagged sound profiles. If the generated sound profile is determined to match another sound profiles, then the automated speech recognition processing may be stopped, and the voice activated electronic device may be instructed to return to a keyword spotting mode. If the matching sound profile is not already stored in a database of known sound profiles, it can be stored for future comparisons.

Claims20

Dependent claims

A method, comprising:

receiving, at a backend system, first audio data;
receiving a first timestamp indicating a first time that the first audio data was sent to the backend system by a first user device;
receiving, at the backend system, second audio data;
receiving a second timestamp indicating a second time that the second audio data was sent to the backend system by a second user device;
determining that an amount of time between the first time and the second time is less than a predetermined period of time, which indicates that the first audio data and the second audio data were sent at a substantially same time;
generating a first audio fingerprint of the first audio data by performing a first fast Fourier transform (“FFT”) on the first audio data, the first audio fingerprint comprising first data representing a first time-frequency profile of the first audio data;
generating a second audio fingerprint of the second audio data by performing a second FFT on the second audio data, the second audio fingerprint comprising second data representing a second time-frequency profile of the second audio data;
determining a bit error rate between the first audio fingerprint and the second audio fingerprint by determining a number of different bits between the first audio fingerprint and the second audio fingerprint, and then dividing the number by a total number of bits;
determining that the bit error rate is less than a predefined bit error rate threshold value indicating that the first audio data and the second audio data both represent a same sound;
and
storing the first audio fingerprint as a flagged audio fingerprint in memory on the backend system such that receipt of additional audio data that has a matching audio fingerprint is ignored by the backend system.

8 more elements

A backend system, comprising:

memory;
communications circuitry;
and
at least one processor operable to:
receive first audio data;
receive a first timestamp indicating a first time that the first audio data was sent to the backend system by a first user device;
receive second audio data;
receive a second time stamp indicating a second time that the second audio data was sent to the backend system by a second user device;
determine that an amount of time between the first time and the second time is less than a predetermined period of time, which indicates that the first audio data and the second audio data were sent at a substantially same time;
generate a first audio fingerprint of the first audio data by performing a first fast Fourier transform (“FFT”) on the first audio data, the first audio fingerprint comprising first data representing a first time-frequency profile of the first audio data;
generate a second audio fingerprint of the second audio data by performing a second FFT on the second audio data, the second audio fingerprint comprising second data representing a second time-frequency profile of the second audio data;
determine a bit error rate between the first audio fingerprint and the second audio fingerprint by determining a number of different bits between the first audio fingerprint and the second audio fingerprint, and then dividing the number by a total number of bits;
determine that the bit error rate is less than a predefined bit error rate threshold value indicating that the first audio data and the second audio data both represent a same sound;
and
store the first audio fingerprint as a flagged audio fingerprint in the memory such that receipt of additional audio data that has a matching audio fingerprint is ignored.

12 more elements

Dependencies are inferred from the claim text.

Facts

Application: 15/195,587Filed 28 Jun 2016
Granted: 8 Aug 2017
Term ends: Not estimatedThe source records don’t support an estimate.

Classification6 codes

G10L 15/22
Speech recognition: procedures used during a speech recognition process, e.g. man-machine dialogue
G10L 25/18
Characterised by the type of extracted parameters: the extracted parameters being spectral information of each sub-band
G10L 25/51
Specially adapted for particular use: for comparison or discrimination
G10L 19/08
Using predictive techniques: determination or coding of the excitation function; Determination or coding of the long-term prediction parameters