18328343. SYSTEM AND METHOD FOR POST-ASR FALSE WAKE-UP SUPPRESSION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

SYSTEM AND METHOD FOR POST-ASR FALSE WAKE-UP SUPPRESSION

Organization Name

SAMSUNG ELECTRONICS CO., LTD.

Inventor(s)

Tapas Kanungo of Redmond WA (US)

Preeti Saraswat of Santa Clara CA (US)

Stephen Michael Walsh of Sunnyvale CA (US)

SYSTEM AND METHOD FOR POST-ASR FALSE WAKE-UP SUPPRESSION - A simplified explanation of the abstract

This abstract first appeared for US patent application 18328343 titled 'SYSTEM AND METHOD FOR POST-ASR FALSE WAKE-UP SUPPRESSION

Simplified Explanation

The method described in the abstract involves using machine learning models to predict the likelihood of a wake word or phrase being spoken in a speech signal, and then taking action based on that prediction. Here is a simplified explanation of the abstract:

• Obtaining a speech signal
• Predicting the likelihood of a wake word or phrase using machine learning models
• Performing automatic speech recognition if the likelihood exceeds a threshold
• Generating instructions to perform an action based on the prediction

Potential Applications

This technology could be applied in smart speakers, virtual assistants, and other voice-controlled devices to improve wake word detection and response accuracy.

Problems Solved

This technology solves the problem of accurately detecting wake words or phrases in speech signals, leading to more reliable and efficient voice command recognition systems.

Benefits

The benefits of this technology include improved user experience with voice-controlled devices, faster response times to voice commands, and increased overall efficiency in performing actions based on spoken instructions.

Potential Commercial Applications

A potential commercial application of this technology could be in the development of advanced voice-controlled devices for smart homes, offices, and other environments where hands-free operation is desired.

Possible Prior Art

One possible prior art for this technology could be existing speech recognition systems used in virtual assistants and smart speakers, which may also use machine learning models for wake word detection and response.

How does this technology handle different accents and speech patterns?

This article does not address how the machine learning models are trained to account for variations in accents and speech patterns that may affect wake word detection accuracy.

What measures are in place to protect user privacy and data security in this technology?

The article does not discuss any specific privacy or security features implemented in this technology to ensure the protection of user data and sensitive information.

Original Abstract Submitted

A method includes obtaining a speech signal. The method also includes predicting a first likelihood of a wake word or phrase being spoken in the speech signal using a first machine learning model trained to receive the speech signal as input. The method further includes, responsive to the first likelihood exceeding a first threshold, performing automatic speech recognition on the speech signal to determine a textual representation of the speech signal. The method also includes predicting a second likelihood of the wake word or phrase being spoken in the speech signal using a second machine learning model trained to receive at least one of the textual representation, audio features associated with the speech signal, and context features associated with the electronic device. In addition, the method includes, responsive to the second likelihood exceeding a second threshold, generating instructions to perform an action requested in the speech signal.