nfxbeats/sonicdna

SonicDNA is a desktop app for musicians and producers to find audio samples that sound similar to one another. You give it a sample, point it at a folder full of audio files, and it returns a ranked list of the closest matches based on qualities like pitch, brightness, attack, and tone rather than just the file name. Everything runs locally on your own computer. Nothing is uploaded anywhere, and the app never changes your original audio files. It builds a small index that stores file paths and extracted sound characteristics, but not copies of the audio itself. The application works on Windows, macOS, and Linux, and needs Python 3.12 or newer. A launcher script sets up an isolated Python environment and installs everything the first time you run it, which can take a few minutes. After that, you can run it either through the graphical interface, where you add folders to scan, drop in a sample, and click a button to see matches, or from the command line if you prefer typing a single command with a query file and a library folder. The interface includes a few practical touches: you can preview the query sound and each search result with a click, drag matching files straight into music software, sort results by similarity score, export the results as a spreadsheet-friendly file, and adjust how much weight is given to different sound qualities like pitch or brightness when ranking matches. It also remembers your settings and folders between sessions, and supports light, dark, and custom color themes. Because indexing is incremental, SonicDNA only processes new or changed files on later scans rather than starting from scratch each time, unless you specifically ask it to rebuild everything.