Audio sample management

2021-09-14 — 2022-02-19

Wherein audio samples are catalogued by machine-listening algorithms, and 2D similarity maps are employed to surface drum hits and foster serendipitous discovery within production workflows.

computers are awful
content
data sets
diy
machine listening
making things
music
search
signal processing
Figure 1: Put a banging donk on it

Here are sample library indexers that use machine listening to classify audio for music production purposes.

1 Atlas 2

Atlas 2 (USD99) (Windows/macos/linux)

I have, use, and love this. It is a combination drum sequencer, sampler, and indexer. The interface is smooth and well-designed but definitely quirky. Undo is limited, but there is much scope for random re-dos. The similarity search seems powerful, but most of the information is hidden from the user in favour of a shiny animated visualization through which we can browse.

Figure 2

It looks like some kind of classic 2d clustering such as UMap driving the search, which is fine IMO; that is more or less what those algorithms were designed for. You are encouraged to lean into serendipity to get stuff done. I would say that this attempts to surf serendipity, which is probably the right approach for me personally.

2 Sononym

Sononym (USD99) (Windows/macos/linux) is more of an indexer and less of a sampler in its own right. It also focuses on drum sounds but has a richer vocabulary for them than Atlas; it seems to be designed to supplement traditional workflows with precision searching rather than as an instrument in its own right.

Since I already seem to suffer some kind of OCD when it comes to sample organization, I am restricting myself to Atlas rather than get lost in the excessive informativeness of Sononym. This is probably about my brain.

3 Incoming

XO - XLN Audio

Looks like Atlas 2 but…

  • better sequencer
  • no linux support

Polarity’s intro video is pretty good I didn’t know xo could do that? : Bitwig

4 Acquiring samples

That’s a different thing; see sample libraries.