Patent · US Active

Speech recognition system providing seclusion for private speech transcription and private data retrieval

US11031010B2 · kind B2 · utility

2Cited by
0References
15Claims
0Family size

Assignee

Inventors

Key dates

Filing dateJul 3, 2019
Grant dateJun 8, 2021
Priority date
Expiry dateDec 16, 2039

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG10L2015/227
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

A method includes receiving a voice input via a microphone of an electronic device, and determining whether the voice input contains speech from an authorized user of the electronic device or speech from an unauthorized user. The method includes in response to determining that the voice input contains speech from the authorized user: determining whether the speech contains private speech or public speech; in response to determining that the speech contains private speech, processing the voice input through a local automatic speech recognition (ASR) engine within the electronic device, the local ASR engine converting the voice input from audio format to text format and outputting a text transcription of the private speech; and in response to determining that the speech does not contain private speech, forwarding the voice input through a communication interface associated with a network-connected external device for processing the voice input at the network-connected external device.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.