This feature is created in the context of the Dataset Creation Toolkit GSoC project. It's related to data validation during the dataset creation/submission process. Once new low-level data are extracted from dataset items and submitted to the server the low-level->metadata->audio properties->md5_encoded is used to search for duplicates in the acousticbrainz.lowlevel_json table. The main changes are related to the acousticbrainz-server/db/data.py and relative test file.