The technology developed by Advestigo uses different techniques to extract content information from different media types, or Elementary Media - Text, Image, Sound.
Audio fingerprints are extracted from multiple timescale slices of the signal, to identify different characteristics of the content. By taking timeslice fingerprints, the analysis is capable of detecting partial copies or sampled and remixed audio. The analysis is also able to handle alterations such as changes in volume or signal-to-noise ratio.
Examples of robust audio analysis: - Additions, insertions and deletions of parts of the audio stream, such as remixes or talkovers
- Extraneous background noise, echo, changes in volume
- Clipping: partial cuts, late intro, overlap with preceding recordings on the same media
- Sampling and/or lossy compression, for example MP3 encoding at different levels of compression.
|
