US 10,950,249 B2Granted

Audio watermark encoding/decoding

AssigneeAmazon Technologies, Inc.

InventorsYuan-Yen Tai, Mohamed Mansour, Parind Shah

Life of this patent

Estimating the term…

Loading patent…

Abstract

A system may embed audio watermarks in audio data using an Eigenvector matrix. The system may detect audio watermarks in audio data despite the effects of reverberation. For example, the system may embed multiple repetitions of an audio watermark before generating output audio using loudspeaker(s). To detect the audio watermark in audio data generated by a microphone, the system may perform a self-correlation that indicates where the audio watermark is repeated. In some examples, the system may encode the audio watermark using multiple repetitions of a multi-segment Eigenvector. Additionally or alternatively, the system may encode the audio watermark using a binary sequence of positive and negative values, which may be used as a shared key for encoding/decoding the audio watermark. The audio watermark can be embedded in output audio data to enable wakeword suppression (e.g., avoid cross-talk between devices) and/or local signal transmission between devices in proximity to each other.

Claims23

Dependent claims

A computer-implemented method, the method comprising, by at least one server:

receiving a first instruction to embed an audio watermark in first audio data;
determining a shared key associated with the audio watermark, the shared key corresponding to a sequence of positive values and negative values and including:
a first sign associated with a first segment of a first portion of the first audio data,
a second sign associated with a second segment of the first portion of the first audio data,
a third sign associated with a first segment of a second portion of the first audio data, and
a fourth sign associated with a second segment of the second portion of the first audio data;
determining a first identifier associated with the first audio data;
storing an association between the first identifier and the shared key;
determining a first value corresponding to a size of a frequency range within which to embed the audio watermark;
generating a first Eigenvector, the first Eigenvector having a length equal to the first value;
generating a second Eigenvector based on the first Eigenvector;
generating first data by multiplying the first sign and the second Eigenvector;
generating second data by multiplying the second sign and a third Eigenvector;
generating third data by multiplying the third sign and the second Eigenvector;
generating fourth data by multiplying the fourth sign and the third Eigenvector;
generating a first portion of second audio data by inserting the first data within the first segment of the first portion of the first audio data;
generating a second portion of the second audio data by inserting the second data within the second segment of the first portion of the first audio data;
generating a third portion of the second audio data by inserting the third data within the first segment of the second portion of the first audio data;
generating a fourth portion of the second audio data by inserting the fourth data within the second segment of the second portion of the first audio data;
generating the second audio data by combining the first portion of the second audio data, the second portion of the second audio data, the third portion of the second audio data, and the fourth portion of the second audio data;
sending, to a first device, the second audio data;
and
sending, to the first device, a first command instructing the first device to generate, via at least one loudspeaker of the first device, output audio based on the second audio data, wherein the audio watermark is represented in the output audio.

20 more elements

A computer-implemented method, the method comprising:

receiving a first instruction to embed an audio watermark in first audio data;
determining a first identifier associated with the first audio data;
determining a sequence of parameters associated with the audio watermark, the sequence corresponding to a first value and a second value, wherein the first value offsets the second value;
storing an association between the first identifier and the sequence;
determining a third value corresponding to a size of a frequency range within which to embed the audio watermark;
generating a first Eigenvector, the first Eigenvector having a length equal to the third value;
generating a second Eigenvector based on the first Eigenvector;
generating a first portion of second audio data by inserting, within a first portion of the first audio data, first data corresponding to the second Eigenvector and the sequence;
generating a second portion of the second audio data by inserting, within a second portion of the first audio data, second data corresponding to a third Eigenvector and the sequence;
and
generating the second audio data by combining the first portion of the second audio data and the second portion of the second audio data.

8 more elements

A system comprising:

at least one processor;
and
memory including instructions operable to be executed by the at least one processor to cause the system to:
receive a first instruction to embed an audio watermark in first audio data;
determine a first identifier associated with the first audio data;
determine a sequence of parameters associated with the audio watermark, the sequence corresponding to a first value and a second value, wherein the first value offsets the second value;
store an association between the first identifier and the sequence;
determine a third value corresponding to a size of a frequency range within which to embed the audio watermark;
generate a first Eigenvector, the first Eigenvector having a length equal to the third value;
generate a second Eigenvector based on the first Eigenvector;
generate a first portion of second audio data by inserting, within a first portion of the first audio data, first data corresponding to the second Eigenvector and the sequence;
generate a second portion of the second audio data by inserting, within a second portion of the first audio data, second data corresponding to a third Eigenvector and the sequence;
and
generate the second audio data by combining the first portion of the second audio data and the second portion of the second audio data.

11 more elements

Dependencies are inferred from the claim text.

Facts

Application: 16/141,489Filed 25 Sep 2018
Published: US 2020/0098379 A126 Mar 2020
Granted: 16 Mar 2021
Term ends: Not estimatedThe source records don’t support an estimate.

Classification9 codes

G10L 19/018
Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals,…:…
G06F 3/167
Sound input; Sound output: audio in a user interface, e.g. using voice commands for navigating, audio feedback
G10L 13/08
Speech synthesis; Text to speech systems: text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme…
G10L 15/05
Segmentation; Word boundary detection: word boundary detection