If you've pitched sync for any length of time, you know this type of brief:
That brief contains no BPM. No key. No genre tag. Just an artist reference and a vibe description, and you have 24 hours to respond with your best songs.
This is the hardest kind of brief to match manually. You can't filter a spreadsheet for "Harry Styles vibes." You can't type that into Disco's search bar. You have to translate the reference yourself, hold it in your head, and then go find songs in your catalog that match a description you've just invented.
With a 50-song catalog, this is manageable. With 500+ songs, you're going to miss things.
When a music supervisor writes "sounds like Harry Styles' Fine Line era," they're encoding a specific set of sonic and thematic descriptors. A human ear unpacks that automatically. AI sync music software has to do the same thing programmatically.
For the Fine Line reference, the descriptors include: mid-tempo (95-115 BPM), warm analog production, guitar-forward with vintage synth texture, emotionally open and introspective lyrics, themes of vulnerability and self-examination, indie-pop with classic rock influences, high production polish.
Here's how a few common sync brief references decode:
The problem with doing this decoding yourself, beyond the cognitive load, is consistency. When you're under deadline, under caffeine, and trying to remember 600 songs at once, your mental model of what each song sounds like gets fuzzy. You pitch the songs you remember most vividly, not the songs that best fit the decoded descriptors.
The songs you wrote two years ago that are perfect for a brief you received today exist in your catalog. You just can't reliably find them in 24 hours without a system.
The approach AI-powered sync music software takes is to pre-index your catalog against the same descriptors that artist references map to. When a brief with an artist reference comes in, the software decodes the reference into those descriptors and scores every song in your catalog against them.
The result: your top 10 best-matching songs, ranked, in seconds. Not the songs you remembered. The songs that actually fit.
For a 500-song catalog receiving 3-5 briefs per week, the working sync songwriter's reality, this changes how many pitches you can respond to, how fast you can respond, and how accurate each pitch is.
Pitchkit indexes your first 30 songs free, the whole Mac app, every feature, no account and no card. Paste any brief and it ranks your catalog by fit. $39 once removes the cap.
Download free trial: 30 songsPartially, and it helps to know which part. A reference like 'something in the Phoebe Bridgers lane' can be translated into concrete musical characteristics, tempo range, instrumentation, vocal register, production space, lyrical register, and those characteristics can then be matched against a catalog that has been analysed for the same attributes.
What software can't do is know that a particular supervisor uses an artist name to mean something idiosyncratic, or that a brand brief has an unstated ceiling on how aggressive the track can be. It narrows nine hundred songs to fifteen candidates; your judgement picks the five.
It translates the reference into musical characteristics before matching, rather than string-matching the artist's name against your tags. Out of the box it recognises roughly 200 artists from a built-in list. Adding your own Anthropic or Gemini API key in Setup extends that to effectively any artist, including very new or very obscure ones.
The only data that ever leaves your Mac in that mode is the artist name itself, never your audio, lyrics or catalog.
No. Everything works without one: audio analysis, tempo and mood matching, lyric transcription, pitch folders, and artist references against the built-in list of around 200 names.
An API key only widens artist-reference matching beyond that list. If your briefs mostly use mainstream references you may never need one; if they name deep-cut or very current artists, it's worth adding.