Milos Miljkovic: Song Matching by Analyzing and Hashing Audio Fingerprints

PyData NYC 2015 We shall dive into the science of song matching using audio analysis and search algorithms in a database store. Python's rich scientific stack will be used for sound processing and hashing, and SQL database will serve for storing of audio fingerprints and matching unknown songs. Audio analysis coupled to SQL database store will serve as the basis of this talk. The audience will be introduced to the following concepts: Nature of sound. Digital representation of sound. Short-time Fourier transform and spectrograms. Analysis of spectrograms in order to produce audio fingerprints. Hashing of audio fingerprints. Building of SQL database to store song audio hashes and meta-data. Searching of DB for matching unknown songs. Code for scraping YouTube for songs used in database building will also be provided to make following of the talk easier and provide a basis for the exploration of audio processing and DB creation. Slides available here: https://github.com/miishke/PyDataNYC2015 00:00 Welcome! 00:10 Help us add time stamps or captions to this video! See the description for details. Want to help add timestamps to our YouTube videos to help with discoverability? Find out more here: https://github.com/numfocus/YouTubeVi...