×

Methods and systems for performing signal analysis to identify content types

  • US 9,653,094 B2
  • Filed: 04/21/2016
  • Issued: 05/16/2017
  • Est. Priority Date: 04/24/2015
  • Status: Active Grant
First Claim
Patent Images

1. A method of processing audio signals to identify content, the method comprising:

  • receiving digitized audio content;

    decoding the audio content using a decoder;

    segmenting frames of the decoded audio content by applying a windowing function to a given audio frame using a first window type having a time width approximately equal to a delay time of the decoder;

    calculating an estimate of a power spectrum of a given frame;

    applying a mel filter bank to the power spectrum of the given frame and providing resulting filter bank energies;

    applying a DCT matrix to the resulting filter bank energies to generate a DCT output;

    taking a log of the DCT output to generate a mel coefficient 1;

    dynamically calculating a first threshold for the content; and

    utilizing the mel coefficient 1 and the dynamically calculated first threshold to detect a near silence between content of different types and to identify the types of content separated by the near silence.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×