Techniques are provided for implementing multi-track audio analysis. In some instances, a plurality of audio tracks are received and combined into a single audio file. A collection of audio file characteristics are extracted from the single audio file. In some examples, the collection of audio file characteristics are associated with of the plurality of audio tracks and the single audio file. Audio characteristic edits are received for revising the collection of audio file characteristics. Metadata associated with each of the plurality of audio tracks and for the single audio file are updated based at least in part on the audio characteristic edits.