17959637. GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S) simplified abstract (GOOGLE LLC)
Contents
- 1 GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S)
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S) - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Original Abstract Submitted
GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S)
Organization Name
Inventor(s)
Rajiv Mathews of Sunnyvale CA (US)
Dragan Zivkovic of Sunnyvale CA (US)
Khe Chai Sim of Dublin CA (US)
GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S) - A simplified explanation of the abstract
This abstract first appeared for US patent application 17959637 titled 'GENERATION AND UTILIZATION OF PSEUDO-CORRECTION(S) TO PREVENT FORGETTING OF PERSONALIZED ON-DEVICE AUTOMATIC SPEECH RECOGNITION (ASR) MODEL(S)
Simplified Explanation
The abstract describes a patent application related to on-device processors storing corrections for Automatic Speech Recognition (ASR) processing of audio data, as well as personalizing ASR models based on these corrections.
- On-device processors store corrections for ASR processing in on-device storage with a time to live (TTL).
- Corrections include modified speech hypotheses and are used to personalize on-device ASR models.
- Pseudo-corrections are stored based on additional ASR processing to prevent forgetting by the ASR model.
Potential Applications
This technology could be applied in various industries such as healthcare, customer service, and education for improving speech recognition accuracy and personalizing user experiences.
Problems Solved
1. Enhanced accuracy: By storing corrections and pseudo-corrections, the ASR model can continuously improve and adapt to individual speech patterns. 2. Preventing forgetting: Personalizing the ASR model based on corrections helps prevent the model from forgetting specific speech patterns over time.
Benefits
1. Improved user experience: Personalized ASR models can better understand and transcribe user speech, leading to more accurate results. 2. Continuous learning: By storing corrections and pseudo-corrections, the ASR model can continuously learn and adapt to new speech patterns.
Potential Commercial Applications
"Enhancing Speech Recognition Accuracy and Personalization in Client Devices" could be used in smart speakers, smartphones, and other devices with ASR capabilities to provide more accurate and personalized speech recognition services.
Possible Prior Art
One potential prior art could be the use of machine learning algorithms to personalize ASR models based on user corrections. Another could be the storage of speech corrections in on-device storage for improving ASR accuracy over time.
Unanswered Questions
The patent abstract does not mention any specific methods or technologies used to address data privacy concerns related to storing speech corrections on the device. This could be a potential area of concern for users and regulators, and it would be important for the patent to outline how data privacy is maintained in this process.
What is the impact of storing corrections and pseudo-corrections on device storage capacity and performance?
The abstract does not provide information on how storing corrections and pseudo-corrections may impact device storage capacity and performance. It would be important to understand the potential implications of this technology on device resources to assess its feasibility and scalability in real-world applications.
Original Abstract Submitted
On-device processor(s) of a client device may store, in on-device storage and in association with a time to live (TTL) in the on-device storage, a correction directed to ASR processing of audio data. The correction may include a portion of a given speech hypothesis that was modified to an alternate speech hypothesis. Further, the on-device processor(s) may cause an on-device ASR model to be personalized based on the correction. Moreover, and based on additional ASR processing of additional audio data, the on-device processor(s) may store, in the on-device storage and in association with an additional TTL in the on-device storage, a pseudo-correction directed to the additional ASR processing. Accordingly, the on-device processor(s) may cause the on-device ASR model to be personalized based on the pseudo-correction to prevent forgetting by the on-device ASR model.
