redline-v1

AI timing for osu! — a 45M-parameter frame-level model that turns raw audio into osu! timing (redlines: BPM / offset / meter), built for the songs that are hard to time by hand: live drummers, metal, rubato, accelerandos. Output is a starting point to verify by ear.

Code and solver: https://github.com/Tiger14n/Redline

Model

Encoder-only transformer (rotary attention, conv frontend over log-mel + onset flux), per-frame outputs at 50 fps:

head	output
beat	beat-here logit
down	downbeat (measure start) logit
off	sub-frame offset (recovers ~2-5 ms precision from 20 ms frames)
tempo	BPM-bin logits (per-frame tempo curve)
change	major (audible) tempo/meter change logit
red	notation logit — where a human mapper writes a redline, including convention re-anchors

d_model 512, depth 14, 8 heads, FFN 2048. Trained on ranked osu! beatmaps (three-stage lineage; configs in the repo) on a single RTX 3080.

Results (held-out data)

Model heads, 400 held-out mapsets: beat F1 0.955, downbeat F1 0.892 (frame-level, ±40 ms), offset MAE 4.7 ms.

End-to-end (audio → redlines, 366 held-out ranked mapsets, ~50+ per genre, scored against the ranked map's human timing):

genre	beats within 10 ms (avg)	typical (median) map
Pop (55)	88.6 %	96.5 %
Metal (48)	88.1 %	95.7 %
Rock (51)	86.7 %	94.4 %
Electronic (49)	84.2 %	97.1 %
Video game (54)	82.8 %	97.0 %
Anime (55)	79.7 %	95.0 %
other (54)	80.2 %	93.2 %
all (366)	84.2 %	96.0 %

The typical song comes out essentially on-grid (median 93–97 % in every genre); a hard minority (live/rubato, quiet audio) needs hand-finishing. Songs whose human timing is a single redline get a median of exactly 1 redline. Full tables, redline-placement metrics and methodology in the repo README.

Usage

pip install -r requirements.txt   # from the repo
hf download Tiger14n/redline-v1 redline-v1.pt --local-dir models/
python -m redline.generate_timing song.mp3 --osz

The checkpoint contains model (state dict) and args (data representation + architecture), which redline.infer.Model2Beats uses to reconstruct the exact training-time pipeline.