Listening Diary
Listening Diary is a system that logs every song the user plays and scores the mood of the audio itself with a neural audio model (CLAP). A parallel channel scores the mood of the lyrics (multilingual text embeddings), so the diary can surface the most interesting signal of all: when a song sounds one way but reads another (“sounds happy · reads sad”). It can be run entirely on a local machine.
The diary below is a single self-contained, interactive HTML view — a month heatmap of daily mood, a two-lane timeline putting audio next to lyrics, and divergence tags where the two disagree.
Note: this is a demo populated with synthetic data — invented songs and artists, not real listening history.
👉 Open the full interactive demo → (best experience — hover for per-song mood charts, click a busy evening to expand it; these float past the preview box above).