Texas Instruments Corporation

Amazon Technologies, Inc.

Capital One Services, LLC

Bozhao Tan

Jianchun Dong

Xuan Zhong

Chia-Jean Wang

Geng Ye

Han Zhang

Isabelle Alice Yvonne Moulinier

David Almquist

June Wu

Wei Ma

Rohit Joshi

Zhenyong Zhang

Jonathan Weimer

Zhen Xu

Ashish Gupta

Haoxing Zhang

Fedor N Dzegilenko

Jeffrey Michel

Meenakshi Panda

Vinayak Hulawale

Vikash Tiwari

Methods and systems are disclosed herein for improving the quality of audio for use in a biometric. A biometric system may use machine learning to determine whether audio or a portion of the audio should be used as a biometric for a user. A sample of the user's voice may be used to generate a voice signature of the user. Portions of the audio that do not meet a similarity threshold when compared with the voice signature may be removed from the audio. Additionally or alternatively, interfering noises may be detected and removed from the audio to improve the quality of a voice biometric generated from the audio.

Machine learning for improving quality of voice biometrics

In some implementations, a system may obtain a first model that is trained to identify feature data associated with a client system using one or more services of a service platform. The system may train, based on the feature data, a second model to identify anomalies associated with devices accessing the one or more services in association with a client identifier of the client system. The system may receive access data associated with an acting device accessing a service of the service platform. The system may determine, using the second model, that the acting device accessing the service corresponds to potential anomalous activity based on the access information. The system may obtain, from a verification device, a verification that the acting device accessing the service is anomalous activity. The system may perform, based on obtaining the verification, an action associated with the acting device.

Detection of anomalies associated with fraudulent access to a service platform

A head-mounted wearable device provides many functions to a user. The functions may be provided based in part on whether the device is being worn (donned) or not worn (doffed). Force sensing resistors in the device provide output indicative of forces associated with wearing the device. A time series of values indicative of the magnitude of a force may be determined. An increase or decrease in force values measured at different times may indicate that the device has been donned or doffed, respectively. If a subsequent force value does not differ substantially from a previous force value, this may indicate that the device has remained donned or doffed.

Sensors for determining don/doff status of a wearable device

A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. Sparse sound parameter information is extracted from the analog signal. The extracted sparse sound parameter information is processed using a speaker dependent sound signature database stored in the sound recognition sensor to identify sounds or speech contained in the analog signal. The sound signature database may include several user enrollments for a sound command each representing an entire word or multiword phrase. The extracted sparse sound parameter information may be compared to the multiple user enrolled signatures using cosine distance, Euclidean distance, correlation distance, etc., for example.

User programmable voice command recognition based on sparse features

A user interface for a head-mounted wearable device provides many functions to a user. Information about whether the device is being worn (donned) or not worn (doffed) may be acquired in several ways. Force sensing resistors in the device may provide output indicative of the pressure associated with wearing the device. Output from a bone conduction (BC) microphone may be analyzed to determine if the device is in use. Piezoelectric BC speakers typically used to present audio output to a wearer provide changes in voltage corresponding to a change in pressure associated with wearing the device. A BC speaker may emit a signal that is detected by the BC microphone, with changes in the signal strength being used to determine if the device is donned or doffed.

Wearable device don/doff sensor

A head-mounted wearable device (HMWD) may incorporate bone conduction speakers (BCS) to generate audio output that is perceptible to a wearer. During operation, vibrations of the BCS may produce sound in the surrounding air that is perceptible to bystanders. An active noise cancellation module monitors the output and generates cancellation audio that is out of phase with the sound leaked by the BCS. An air conduction speaker emits the cancellation audio, producing destructive acoustic interference to the leaked sound. As a result, the user of the HMWD is able to hear audio clearly while bystanders are not.

Active noise cancellation for bone conduction speaker of a head-mounted wearable device

A head-mounted wearable device (HMWD) may incorporate bone conduction speakers (BCS) to generate audio output that is perceptible to a wearer. Due to the mechanical connection between the BCS and the rest of the device, when the output amplitude of the BCSs is too great, the air-conducted audio 'leaks' and may be heard by other people. This leakage may result in eavesdropping by, or generate a nuisance to, those other people. The noise level in the environment around the HMWD is determined and used to dynamically adjust the amplitude of the audio output generated by the BCS. This adjustment improves the comfort and intelligibility of the audio output to the wearer while minimizing audio leakage to the surrounding environment.

Dynamic range management of bone conduction speaker output

A head-mounted wearable device incorporates a transducer that operates as a bone conduction (BC) microphone. Vibrations from a user's speech are transferred through the head of the user to the BC microphone. An air conduction (AC) microphone detects sound transferred via air. Signals from the BC microphone and the AC microphone are compared to determine if a common signal is present in both. For example, both signals may have a cross-correlation that exceeds a threshold value. Based on the comparison, voice activity data is generated that indicates the user wearing the device is speaking.

Voice activity detection using air conduction and bone conduction microphones

A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. Sparse sound parameter information is extracted from the analog signal. The extracted sparse sound parameter information is processed using a sound signature database stored in the sound recognition sensor to identify sounds or speech contained in the analog signal, wherein the sound signature database comprises a plurality of sound signatures each representing an entire word or multiword phrase.

Acoustic sound signature detection based on sparse features

A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. The received analog signal is evaluated using a detection portion of the analog section to determine when background noise on the analog signal is exceeded. A feature extraction portion of the analog section is triggered to extract sparse sound parameter information from the analog signal when the background noise is exceeded. An initial truncated portion of the sound parameter information is compared to a truncated sound parameter database stored locally with the sound recognition sensor to detect when there is a likelihood that the expected sound is being received in the analog signal. A trigger signal is generated to trigger classification logic when the likelihood that the expected sound is being received exceeds a threshold value.

Powering-up AFE and microcontroller after comparing analog and truncated sounds

A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. Sparse sound parameter information is extracted from the analog signal. The extracted sound parameter information is sampled in a periodic manner and a context value is updated to indicate a current environmental condition. The sparse sound parameter information is compared to both the context value and a signature sound parameter database stored locally with the sound recognition sensor to identify sounds or speech contained in the analog signal, such that identification of sound or speech is adaptive to the current environmental condition.

Growing community of inventors

Sunnyvale, CA, United States of America

Bozhao Tan