#1July 2026 · full-stack · deployed
Stemline
Version control for unreleased music — stem-level commits, audio-aware diffs, E2E encryption
Git for audio: commit individual stems, fork lineages without duplicating storage, and hear what changed with A/B crossfade and phase-cancellation diffs. Mixdowns split into stems on-device via HT-Demucs (onnxruntime-web / WebGPU→WASM) so audio never leaves the machine during inference. Content-addressed FastCDC uploads with AES-GCM per chunk; only changed stem bytes go up. Sample-locked multi-stem mixer, waveform comments, and RSA-OAEP wrapped project shares — storage only ever holds ciphertext.








