Splitsmith Pipeline Quickstart GitHub
Project Register Vol. 01 · Ed. 04

Splits from your stage video.

Shot timers like the CED7000 give you splits during practice — but at matches you can't carry one. Head-mounted cameras capture audio that contains all the shot information. Splitsmith extracts it and turns it into actionable training data: per-shot times, a CSV of splits, and an FCPXML timeline with frame-aligned markers you can pop open in Final Cut Pro and review on M / Shift+M.

01

Pipeline

Step 01

Ingest

Drop a folder of head-cam clips. The engine auto-matches them to stages by file timestamp.

Step 02

Beep

Auto-snap to the start beep on each stage. Low-confidence detections land in a review queue.

Step 03

Audit

Waveform + per-shot markers from a 3-voter ensemble. Click a marker to inspect votes. Drag to fine-tune.

Step 04

Export

Per-stage or whole-match FCPXML. Open in Final Cut Pro, navigate with M / Shift+M.

02

What goes in. What comes out.

Inputs

Raw match material

  • .MP4Head-mounted cam footage (e.g. Insta360 Go 3S)
  • .JSONStage timing data from SSI Scoreboard
Outputs · per stage

Reviewable training data

  • _trimmed.mp4Lossless cut around the start beep
  • _splits.csvPer-shot splits — the editable source of truth
  • .fcpxmlFinal Cut Pro timeline with frame-aligned markers
  • _report.txtAnomaly report — overshoots, echoes, low confidence
03

Quickstart

# Clone, sync, build the UI $ git clone https://github.com/mandakan/splitsmith.git $ cd splitsmith $ uv sync $ (cd src/splitsmith/ui_static && pnpm install && pnpm build) # Run the full ingest → audit → export workflow $ uv run splitsmith ui --project ~/matches/your-match