| CPC G10L 15/26 (2013.01) [G10L 15/1815 (2013.01)] | 17 Claims |

|
1. A method comprising:
detecting a first event with use of a sensor, the first event to occur while audio data is output on a computing device, the audio data including speech;
generating a transcript of the audio data using a speech-to-text engine;
identifying a first location in the transcript of the speech from the audio data based on a signal from the sensor that detected an occurrence of the first event;
identifying a portion of the transcript that includes the first location based on one or more timestamps that define the portion of the transcript relative to the first location;
generating a link to audio that is associated with the extracted portion of the transcript; and
providing the extracted portion of the transcript and the link to an application;
wherein the identified portion of the transcript starts at the first location, and an extracted portion includes a semantically-continuous block of sentences, which the first location is part of, the semantically-continuous block of sentences including a sentence that is being output during detection of the first event.
|