v0.1.0

Initial release of Presto — audio fingerprinting and recognition in Go.

Highlights

• Dual-algorithm support — constellation peak-pair hashing (Wang 2003) and sub-band mel-band energy comparison (Haitsma-Kalker 2002) behind a strategy interface
• Persistent library — mmap'd binary store format with zero-copy access and algorithm auto-detection
• HTTP server — POST /v1/match, /v1/stats, /healthz, /readyz, /metrics
• Spectral analysis — presto analyze generates annotated spectrogram PNGs with viridis colormap, peak overlays, and fingerprint stats
• Kubernetes-ready — Dockerfile + deploy manifests with security-hardened pod spec
• Zero external Go deps — FFT, WAV I/O, window functions, and inverted-index matcher all from scratch

Performance

Library size	Match time
50 songs	0.41 ms
100 songs	0.74 ms
500 songs	2.64 ms

Install

go install github.com/nddq/presto/cmd/presto@v0.1.0