Upload one isolated sound of up to 20 seconds to Audial's Resynth tab with Serum 2 as the target, and you get back a .SerumPreset that opens in Xfer Serum 2 with its oscillators, filters, envelopes and effects set to reproduce the sound, plus a render and a similarity score.
$10 a month or $100 a year. Unlimited runs, cancel any time. 2 free runs to try it first.
Audial's Serum 2 preset generator takes a sound and gives you the patch. Upload one isolated sound of up to 20 seconds, choose Serum 2 as the target, and the hosted engine returns a .SerumPreset file whose oscillators, filters, envelopes and effects are set to reproduce what you sent it.
No plugin download is needed. The preset opens in Xfer's Serum 2 itself: drop the file into the user presets folder Serum's preset browser points at, or drag it onto the Serum 2 window. Because it is a real preset and not a sample, nothing about it is locked — slow the attack, pull the filter envelope down, add an LFO, play it across the keyboard.
Every run is scored. Alongside the preset you get a render of the patch, two keyboard renders and a JSON report with the similarity score between the patch and your source, the same number shown on the page. Jobs take about 2½ minutes, plus up to 2½ minutes if a machine has to start.
What you get
- Any single sound up to 20 seconds becomes an editable .SerumPreset
- No plugin to install: the preset opens in Xfer Serum 2 itself
- Needs Serum 2.1 or newer; the file is saved under your sample's name
- Returns the preset, a render, two keyboard renders and a similarity score
- Same engine also targets Vital, with the free Audial Synth plugin
- About 2½ minutes a job, plus up to 2½ minutes if a machine has to start
Hear it
The target is a synth pluck made for this demo. The render is the returned preset at a similarity score of 0.987.
Audial vs doing it by hand
| Task | Audial | By hand |
|---|---|---|
| Getting close to the sound | Audial: The engine searches Serum 2's parameters and scores the match against your clip | By hand: Sound design by ear: pick the oscillators, shape the filter and envelopes, compare, repeat |
| Time | Audial: About 2½ minutes a run, plus up to 2½ minutes if a machine has to start | By hand: Minutes for a simple pluck, much longer for a complex sound |
| What you end up with | Audial: A real .SerumPreset: every parameter editable | By hand: The same, and you learn Serum 2 on the way |
| Sounds Serum 2 cannot make | Audial: Gets as close as Serum 2 can, and the score says how close | By hand: The same limit: you cannot patch what the synth cannot produce |
| Cost | Audial: 2 free runs, then $10 a month with unlimited runs; you need your own Serum 2 | By hand: Free apart from your time; you need your own Serum 2 too |
Use it from code
curl -X POST "https://api.audialmusic.ai/api/functions/run/sound2serum" \
-H "X-API-Key: your_api_key" -H "X-User-ID: your_user_id" \
-F "userId=your_user_id" \
-F "original=@/path/to/clip.wav"import audial
audial.config.set_api_key("your_api_key")
audial.config.set_user_id("your_user_id")
result = audial.sound2serum("clip.wav")
print(result["preset"]) # the .SerumPreset built from the clip
print(result["scores"]["huang_similarity"])# In Claude, Cursor or any MCP client with audial-mcp installed:
"Turn ~/Samples/bell_c3.wav into a Serum 2 preset and tell me the similarity score."Frequently asked questions
Upload the sample on the Resynth tab, choose Serum 2 as the target, and run it. You get back a .SerumPreset file named after your sample. Put it in the folder Serum 2's preset browser points at, or drag it onto the Serum 2 window, and it shows up as a preset you can edit.
No. The preset opens in Serum 2 itself — there is nothing extra to download or authorise. You need Xfer's Serum 2, which Audial does not sell or bundle. (The Vital target is the one with a free Audial plugin, Audial Synth.)
Serum 2.1 or newer. The .SerumPreset format is Serum 2's; Serum 1 presets are a different format and are not produced here.
Every preset is scored against the source clip for similarity, and the number is shown on the page next to the download, so you can see how close a particular patch came before you open it. Clean single sounds score highest. A loop, melody, drum pattern or full mix is not what the engine is for, and the page warns you when a clip sounds like several notes in a row.
One isolated sound up to 20 seconds: a held synth note, a pluck, a bell, a bowed note, a pad stab, a sound-design one-shot. Loops and melodies give a poor preset, so trim one to a single note first. Trim the silence off the front and keep it dry, because reverb left in the clip gets modelled into the patch.
Yes. Post the clip to /functions/run/sound2serum, call audial.sound2serum from the Python SDK, or use the sound2serum tool through the Audial MCP server. All three return the preset plus the renders and the score.
