Looking for a flexible acoustic fingerprint with partial content support

[autarc]

Currently none of the open source acoustic fingerprint implementations I am aware of is capable of providing reliable results for partial content. After Echonest shutdown their service for handling Echoprint requests earlier this year, Chromaprint with AcoustID provides at the time the largest royalty free dataset. Unfortunately the algorithm its based on relies on complete audio files instead of snippets or recordings with background noises. A more flexible solution would be necessary to allow proper content recognition of tracks used in video at the web (e.g. Youtube).