Medallia, Inc.

Jay R Pascarella

David Garrod

A speech redaction engine includes a natural language processing (NLP)-based content redaction module receives an automatic speech recognition (ASR) decoding of a decoded portion of said digitized speech signal and utilizes NLP techniques to determine whether it contains sensitive information that should be redacted, and an ASR confidence-based redaction module that receives a confidence indicator and utilizes said confidence indicator to determine, independent of said NLP-based content redaction module, whether said decoded portion contains one or more word(s) that were recognized with a confidence level that is below a threshold. The speech redaction engine includes means for redacting said decoded portion if the NLP-based content redaction module determines that said portion should be redacted, and means for redacting the one or more word(s) if the ASR confidence-based redaction module determines that the one or more word(s) have the confidence level that is below the threshold.

Use of ASR confidence to improve reliability of automatic audio redaction

A method includes removing sensitive information from an audio recording. The method includes receiving a digitally encoded audio recording; transcoding, if necessary, the audio recording to a pre-defined audio format; identifying periods of voice in/activity in the audio recording; segregating the audio recording into a sequence of separated utterances or words; streaming the sequence of separated utterances or words to an ASR server; receiving, for each streamed utterance or word, an associated ASR decoding; receiving, for each ASR decoding, an associated confidence score indicative of a probability that the associated ASR decoding is correct; identifying for redaction; any ASR decoding that contains sensitive information, and any ASR decoding with a confidence score less than a redaction threshold; and preparing a redacted audio recording by eliminating or masking those word(s) and/or utterance(s) for which an associated ASR decoding was identified for redaction.

Growing community of inventors

Cranberry Township, PA, United States of America

Jay R Pascarella