Video & audio

Music taste from data

Pick background music from charts and measured audio features instead of mood guesswork.

Paid $15 list, $12 until 2026-12-31 · one skill, yours for good. The outline, the steps and the file open after purchase.

Background

One line: BGM taste is screened by data and measured by audio features, not guessed. The Douyin hot-song chart (what the public really takes and puts under videos) is the screen, local librosa features (BPM / key / energy / brightness / harmonicity) are the ruler; the crossing of the two gives a BGM reference library of "what the public really buys *and* what really sounds good", made specifically to cure "monotonous BGM".

Why it is built this way

The order runs collect, score, download, measure, then write rules. Two screens sit in front of any listening judgement: heat and persistence decide who the public actually took, librosa decides what the track sounds like. Screening without measuring teaches brainworm hits, measuring without screening drifts off the chart. --min-view 100000 exists so a tiny sample cannot inflate its like-to-view rate, and it also stops the ranked table from being read as a taste verdict.

The Douyin hot-song chart and the Bilibili music section are stored as separate libraries because they measure different things. The chart reports what people put under their own videos, the music section reports what they watch as music. Douyin's heat is a platform index that only compares inside one chart, so the script passes the rank through instead of recomputing it. Every sample carries platform, chart, heat or views, persistence and collection date, and Douyin audio has no direct link, so the machine track is marked pending while the semantic rules still get written.

Where it bites

• Heat mistaken for taste A high chart rank is treated as a style verdict, so the library fills with tracks that carried videos and sound thin on their own. The red line requires every candidate to clear the five auditory dimensions and the machine track before it enters references/library/.
• Two charts stewed together Rules summarised across the Douyin chart and the Bilibili music section describe two different audiences, and the resulting BPM range fits neither. The hard rule keeps one library per platform and one file per chart, dated, so a rule can always be traced back to the chart it came from.
• Samples with no provenance A track saved without its platform, chart, heat and collection date cannot be re-checked later, and the library turns into a folder of songs nobody can defend. No data provenance means no entry, and the library lives only in references/library/, never in the workspace root.
• The Douyin gap papered over Douyin hot songs have no free audio link, so an analysis run can quietly come back with no features for the highest-velocity samples. Step 3 marks that machine track pending and keeps the semantic track running, which keeps the gap visible instead of filling it with numbers from a different source.

When to use it

Use the interaction data on public social charts (heat / usage count / views / like-to-view rate) — the Douyin hot-song chart, the Bilibili music section and similar public charts — as a screen to pick the BGM samples the public actually votes for, then use local librosa to break out the audio features (BPM / key / energy / brightness / harmonicity), and bank the result into an auditory-aesthetics rule library per platform × chart that serves as the reference anchor when choosing BGM or voiceover for a video. Use when the user wants to improve BGM taste, says the BGM is too monotonous, wants to learn from viral scoring, asks what background music to pick, or needs data backing for BGM before making a video.