Amazon Technologies, Inc.

Gengshen Fu

Shiv Naga Prasad Vitaladevuni

Shuang Wu

Paul McIntyre

Thibaud Senechal

Aaron Challenner

Varun Kumar Nagaraja

Chao Wang

Ming Sun

Zoe Adams

Eli Joshua Fidler

Yixin Gao

Michael Thomas Peterson

Rong Chen

Michael James Rodehorst

Yuriy Mishchenko

Sai Kiran Venkata Subramanya Rupanagudi

Cody Jacques

Huitian Lei

Stefan M Bradstreet

Sree Hari Krishnan Parthasarathi

A device is configured to detect multiple different wakewords. A device may operate a joint encoder that operates on audio data to determine encoded audio data. The device may operate multiple different decoders which process the encoded audio data to determine if a wakeword is detected. Each decoder may correspond to a different wakeword. The decoders may use fewer computing resources than the joint encoder, allowing for the device to more easily perform multiple wakeword processing. Enabling/disabling wakeword(s) may involve the reconfiguring of a wakeword detector to add/remove data for respective decoder(s).

Multiple wakeword detection

A device is configured to detect multiple different wakewords. A device may operate a joint encoder that operates on audio data to determine encoded audio data. The device may operate multiple different decoders which process the encoded audio data to determine if a wakeword is detected. Each decoder may correspond to a different wakeword. The decoders may use fewer computing resources than the joint encoder, allowing for the device to more easily perform multiple wakeword processing. Enabling/disabling wakeword(s) may involve the reconfiguring of a wakeword detector to add/remove data for respective decoder(s). Specific decoders may be activated/deactivated depending on device context, thereby efficiently managing device resources.

Audio detection

Systems and methods for preemptive wakeword detection are disclosed. For example, a first part of a wakeword is detected from audio data representing a user utterance. When this occurs, on-device speech processing is initiated prior to when the entire wakeword is detected. When the entire wakeword is detected, results from the on-device speech processing and/or the audio data is sent to a speech processing system to determine a responsive action to be performed by the device. When the entire wakeword is not detected, on-device processing is canceled and the device refrains from sending the audio data to the speech processing system.

Preemptive wakeword detection

Techniques for using a dynamic wakeword detection threshold are described. A device detects a wakeword in audio data using a first wakeword detection threshold value. Thereafter, the device receives audio including speech. If the device receives the audio within a predetermined duration of time after detecting the previous wakeword, the device attempts to detect a wakeword in second audio data, corresponding to the audio including the speech, using a second, lower wakeword detection threshold value.

Dynamic wakeword detection

A system and method performs wakeword detection and automatic speech recognition using the same acoustic model. A mapping engine maps phones/senones output by the acoustic model to phones/senones corresponding to the wakeword. A hidden Markov model (HMM) may determine that the wakeword is present in audio data; the HMM may have multiple paths for multiple wakewords or may have multiple models. Once the wakeword is detected, ASR is performed using the acoustic model.

Wakeword detection using multi-word model

A system and method performs wakeword detection using a neural network model that includes a recurrent neural network (RNN) for processing variable-length wakewords. To prevent the model from being influenced by non-wakeword speech, multiple instances of the model are created to process audio data, and each instance is configured to use weights determined by training data. The model may instead or in addition be used to process the audio data only when a likelihood that the audio data corresponds to the wakeword is greater than a threshold. The model may process the audio data as represented by groups of acoustic feature vectors; computations for feature vectors common to different groups may be re-used.

Speech processing using a recurrent neural network

A device monitors audio data for a predetermined and/or user-defined wakeword. The device detects an error in detecting the wakeword in the audio data, such as a false-positive detection of the wakeword or a false-negative detection of the wakeword. Upon detecting the error, the device updates a model trained to detect the wakeword to create an updated trained model; the updated trained model reduces or eliminates further errors in detecting the wakeword. Data corresponding to the updated trained model may be collected by a server from a plurality of devices and used to create an updated trained model aggregating the data; this updated trained model may be sent to some or all of the devices.

Wakeword training

Techniques for using a dynamic wakeword detection threshold are described. A server(s) may receive audio data corresponding to an utterance from a device in response to the device detecting a wakeword using a wakeword detection threshold. The server(s) may then determine the device should use a lower wakeword detection threshold for a duration of time. In addition to sending the device output data responsive to the utterance, the server(s) may send the device an instruction to use the lower wakeword detection threshold for the duration of time. Alternatively, the server(s) may train a machine learning model to determine when the device should use a lower wakeword detection threshold. The server(s) may send the trained machine learned model to the device for use at runtime.

Growing community of inventors

Sharon, MA, United States of America

Gengshen Fu