MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202617096847 A) filed by Google LLC on August 11, 2026, for Quantifying Unintended Memorization In Automated Speech Recognition Encoders.
Inventors include Shejwalkar, Virat Vishnu; Chauhan, Geeticka; Chien, Steve; Thakurta, Abhradeep Guha; Thakkar, Om Dipakbhai; and Narayanan, Arun.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: A method (700) includes receiving a training data set including un-transcribed speech utterances (308) that each include audio-only data not paired with any corresponding transcription, and obtaining a plurality of training canary transcriptions (420) each including a predetermined number of words that are out-of- distribution from words of the un-transcribed speech utterances. For each training canary transcription, the method also includes generating, using TTS system (440), a corresponding synthetic training canary speech utterance (332T) that recites the predetermined number of words of the training canary transcription and pre-training an audio encoder (210) on a combination of the un-transcribed speech utterances and the synthetic training canary speech utterances. The method also includes measuring an un-intended memorization of the pre-trained audio encoder based on encoder labels (215) predicted by the pre-trained encoder for the synthetic training canary speech utterances.
Disclaimer: Curated by HT Syndication.
