Earthquake detection pipeline via efficient time series similarity search

I think this is quite similar to how music fingerprinting (Shazam, etc.) works, except it deals with multiple channels.<p>EDIT: Their paper confirms it: &quot;FAST adapts a data mining algorithm, originally designed to identify similar audio clips within large databases&quot;

It&#x27;s surprising that people don&#x27;t rely more on automatic arrival time picking + simple outlier rejection. It&#x27;s so much more efficient than these waveform correlation approaches while being much simpler.