Plain-English sample search

Type what you mean. Hear the best matches.

Search “metallic impact with a long dark tail,” “70s dry snare,” or “124 BPM synth loop in F minor.” Sample separates hard constraints from descriptive intent and searches the audio—not just its filename.

1search field for filenames, descriptions, BPM, key, and instrument intent
40–200supported BPM filter range after safe clamping of explicit values
24major and minor key spellings normalized to canonical pitch classes
descriptive combinations across texture, mood, material, era, and character
Search patterns that carry intent

Write a brief, not a tag list.

Results depend on what exists in your library. Descriptive negatives can steer semantic intent; exact exclusion behavior is not guaranteed for every phrase.

Sample search interface with a natural-language query and ranked waveform results
Actual product interface. The list remains a familiar browser even though retrieval is semantic.

Four kinds of language, one result set.

Strong queries do not need every category. Start with the property you can hear most clearly.

Source

Instrument or event: kick, cello, room tone, footsteps, machine hit, riser, vocal phrase.

Character

Warm, brittle, close, wide, dusty, hollow, wet, intimate, unstable, saturated, restrained.

Music

Explicit BPM, BPM ranges, key, mode, loop versus one-shot, or a request to preserve tempo/key around a reference.

Direction

More organic, less bright, shorter, heavier, softer, or between two known reference sounds.

Filename search still matters.

Semantic search is additive. Exact filenames, pack names, and familiar tokens remain useful when you know them.

What you rememberBest queryWhy
The exact filekshmr_clap_017Filename evidence is the fastest route when the token is known.
The sound, not the nametight layered clap with a bright roomSemantic audio meaning can bridge poor filenames.
The musical slot124 BPM bass loop in F minorStructured tempo/key constraints narrow semantic candidates.
A reference fileFind similar → more organic → lock BPMAudio-to-audio retrieval starts closer than text alone.

A better query in 15 seconds.

If results are vague, add one discriminating property—not a paragraph of conflicting adjectives.

01

Name the event

“Percussion” is wide. “Metallic hand percussion hit” removes entire irrelevant families.

02

Add the unusual thing

“With a papery attack” is more discriminating than five generic mood words.

03

Constrain only what is real

Add BPM or key when your arrangement requires it. Do not over-constrain one-shots with metadata that may not apply.

A strong query is a hypothesis, not a commandment. Audition the top results, then switch to a reference-led search when one candidate is close enough to express the target more precisely than another sentence.

One-time purchase · $129 launch price

Search the sound—not the naming convention.

Sample searches and analyzes locally on Apple Silicon. Your audio stays on your Mac. Use it offline after activation, drag results into any DAW, and request a refund within 14 days if it does not improve your workflow.

Get Sample