Search
Find the sound, not the folder
One box over the whole library. Type an instrument, a key and a tempo and the results narrow as you type, across every pack you have ever bought.
- Clip 1Empty
- Clip 2Empty
- Clip 3Empty
- Clip 4Empty
How it works
Type what you hear. Kick, A minor, 140, and the four you want are on screen.
- Step 01
Type the way you'd say it
"kick a minor 140" is read as three filters, not one string. Each one becomes a chip you can drop, so you can widen the tempo or change the key without retyping the query.
kick a minor 140chips: Drums · Kick · A minor · 138–1421,284 of 1.2M - Step 02
Results carry their evidence
Each row shows the waveform, the detected key and tempo, the pack it came from and where it sits in your folders. You are choosing between sounds, not between filenames.
Kick_Warehouse_Am_128.wavA minor · 128 BPM · Warehouse Vol. 3Samples › Drums › Kicks - Step 03
Ask for more like it
Click More like this on any result and the library is reordered by how close each sample sounds to that one, measured on the audio rather than the name. Drag the one you want straight into the DAW.
more like Kick_Warehouse_Am_128ranked by timbre, decay, fundamentaldrag → Ableton clip slot
What you get
Searches sound, not filenames
Key, tempo and instrument come from the audio, so a file called bounce2.wav is as findable as one that was named properly. The pack that could not be bothered stops being a dead zone.
Filters you can see and drop
Every part of the query becomes a chip. Drop the tempo chip to hear every A minor kick you own; drop the key chip to hear everything at 140. No syntax to learn.
More like this, on the audio
Similarity is measured on timbre, decay and harmonic content, not on tags. It is the fastest way to find the four alternatives to a sample that is nearly right.
Keyboard from open to drop
Slash opens search, arrows move through results, space auditions, and Enter reveals in Finder. It is built to be used mid-session without reaching for the trackpad.
Why it works
The index holds four things for every file: the filename, the detected key, tempo and instrument, the pack it arrived in, and a set of audio features. All of it is written when the file is first read, so searching never touches the audio again and stays instant on a library with a million rows.
A query is parsed before it is matched. Instrument words, key spellings and tempo numbers are pulled out and turned into filters; whatever is left over is matched against filenames and pack names. That is why "kick a minor 140" behaves like three filters rather than a phrase that has to appear somewhere. Tempo is matched as a small range, because a 138 BPM loop is usually a fine answer to a 140 BPM question.
Similarity uses the audio features directly. Each sample is a point in a space built from spectral shape, decay envelope and harmonic content, and More like this returns its nearest neighbours. Two kicks from different packs, named nothing alike, land next to each other because they sound alike — which is the thing you were actually asking about.
Problems it prevents
- BeforeYou remember the sound but not the pack, so you open eleven folders.
- With Prod CrateYou describe the sound and the results are on screen before you finish typing.
- BeforeA pack with unhelpful filenames may as well not be on the drive.
- With Prod CrateKey, tempo and instrument come from the audio, so those files are findable anyway.
- BeforeThe kick is nearly right, so you audition sixty more to find the one that is.
- With Prod CrateMore like this puts its nearest neighbours in front of you, ranked by how close they sound.
- BeforeFinding a sample means leaving the session and browsing.
- With Prod CrateSlash, type, arrow, drag. The search never takes focus away from the arrangement for long.
“Our library is 1.2 million samples across forty years of packs. Search is the first thing that made the old ones usable — half of them were never named in a way anyone could search.”
The rest of Prod Crate
Search questions
01How large a library can it handle?
The index is built for millions of rows and results stay instant as you type. In practice the limit is your library, not the index: a million samples is comfortably inside what it is designed for.
02Does search work offline?
Yes. The index lives on the Mac, so filtering and similarity work with no connection. Auditioning a result that is not downloaded needs the network, exactly as opening it in Finder would.
03What exactly is indexed?
Filename, folder path, the detected key, tempo and instrument, the pack the file arrived in, any tags you added, and a set of audio features used for similarity. The audio itself is never sent anywhere to be searched.
04How does More like this decide what's similar?
It compares audio features — spectral shape, decay envelope, harmonic content — and returns nearest neighbours. It does not look at filenames or packs, which is why it finds the right sample in a pack you had forgotten you owned.
05Can I drag results into my DAW?
Yes. Drag a row straight into an Ableton clip slot, a Logic track or any window that accepts a file. Cloud-only files download during the drag, so what lands in the session is a real file on disk.