Amazon technologies, inc. (20240185845). WAKEWORD DETECTION simplified abstract
Contents
WAKEWORD DETECTION
Organization Name
Inventor(s)
Deepak Yavagal of Mercer Island WA (US)
Ajith Prabhakara of Seattle WA (US)
WAKEWORD DETECTION - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240185845 titled 'WAKEWORD DETECTION
Simplified Explanation
Simplified Explanation: The patent application describes techniques for implementing multiple wakeword detectors on a single device, where a digital signal processor (DSP) detects wakewords in captured speech and a companion application determines if the wakeword was spoken by the user.
- **Key Features and Innovation:**
- Use of a DSP to detect wakewords in captured speech. - Companion application trained with user's speech to verify user-spoken wakewords. - Sending audio data to servers for processing if user-spoken wakeword is detected.
Potential Applications: The technology can be applied in smart devices, virtual assistants, security systems, and other voice-controlled devices where user verification is necessary.
Problems Solved: This technology addresses the need for accurate wakeword detection and user verification in voice-controlled devices to enhance security and user experience.
Benefits: - Improved accuracy in wakeword detection. - Enhanced user privacy and security. - Seamless integration with existing voice-controlled devices.
Commercial Applications: The technology can be utilized in smart home devices, automotive systems, healthcare applications, and various industries requiring voice-controlled user authentication.
Prior Art: There is prior art related to wakeword detection and user verification in voice-controlled devices, but the specific combination of DSP-based detection and companion application verification may be novel.
Frequently Updated Research: Ongoing research in speech recognition, machine learning, and user authentication technologies may impact the development and implementation of this technology.
Unanswered Questions:
- **Question 1:** How does the technology ensure user privacy and data security during the audio data processing?
- The technology may use encryption methods and secure communication protocols to protect user data during processing.
- **Question 2:** What are the potential challenges in implementing this technology across different languages and accents?
- Adapting the wakeword detection and user verification algorithms to different languages and accents may require extensive training data and robust machine learning models.
Original Abstract Submitted
techniques for implementing multiple wakeword detectors on a single device are described. a digital signal processor (dsp) of the device may implement a wakeword detection component to detect when captured speech includes a wakeword. a companion application installed on the device may implement a wakeword detection component trained using speech of a user of the device. if the dsp's wakeword detection component detects a wakeword in speech, the companion application's wakeword detection component may be used to determine whether the wakeword was spoken by the user of the device. if the companion application's wakeword detection component determines the user spoke the wakeword, audio data representing the speech may be sent to at least one server(s) for processing.
