Picture yourself listening to the patter of rain when suddenly you hear a clap of thunder. You had no idea when it would come or what it would sound like, yet you caught it anyway. That is how a new method catches gravitational waves: it hunts for any sharp burst, with no templates at all.
Normally the LIGO detectors search for signals by matching them against millions of precomputed patterns, as if they were listening for a familiar tune. But a star exploding or a neutron star trembling has no exact sheet music. So researchers came up with a wavelet filter: it breaks the data stream into short waves of different lengths and checks which of them are too strong to be ordinary noise.
When two detectors catch similar bursts at almost the same moment, it could be one and the same source. Such coincidences used to be checked with a simple formula, but now a neural network does the job. It learns from examples and decides on its own which pairs are real and which are random. In testing, the method found nearly every black hole merger, and even the real event GW250114, without a single ready-made template.
The event GW250114 was found by the pipeline without templates, even though its network signal-to-noise is around 80 and the false alarm rate is extremely small. The wavelet filter uses ten different bases so that every signal is represented in its most compact form.
The idea of hearing the cosmos without ready-made templates echoes Carl Sagan's novel Contact, where a signal of extraterrestrial origin was detected precisely because it did not resemble the expected noise.