Why BPM and genre don't describe mood
Two tracks at 124 BPM, both tagged House. One is a soft, rolling groove for 10 pm; the other is a full-on floor filler for 1 am. On paper they're identical.
Each number DJs usually filter by misses something:
- BPM tells you speed, not intensity. A half-time beat at 140 BPM can feel calmer than a busy track at 120.
- Genre tags are too broad and inconsistent. "House" covers everything from lounge to peak time.
- Key is a weak hint at best. Minor isn't sad and major isn't happy; plenty of euphoric anthems are in minor keys.
- Loudness barely helps. Most modern masters are loud, so a chill track and a banger can measure about the same.
What you hear as energy is a mix of things: how hard and how often the kick hits, how heavy the bass is, how bright the top end is, how busy the arrangement is, how big the drops and breakdowns are. Mood is the color on top: bright or dark, warm or cold, uplifting or tense. No single number captures that, which is why sorting by mood has always been done by ear.
Mood vs energy: two different axes
Many DJs blend the two, and that's why mood folders turn messy over time. Energy is how hard a track pushes. Mood is how it feels. They're independent:
| Bright mood | Dark mood | |
|---|---|---|
| Low energy | Sunset, lounge, chill | Deep, hypnotic, late night |
| High energy | Euphoric, festival, uplifting | Warehouse, driving, peak-time dark |
A deep, dark roller and a sunny lounge track are both low energy, but you'd never swap one for the other. Decide which axis you lean on most when choosing the next track, and build your labels around it. If you really need both, combine them into a few labels like "Peak Dark" and "Peak Euphoric" instead of creating every possible pair.
A starter set of mood and energy labels
There's no standard list. These are labels working DJs commonly use. Take what fits your sets, rename freely, and drop the rest.
| Label | What it usually sounds like | When it usually gets played |
|---|---|---|
| Chill | Soft or no kick, warm pads, lots of space | Dinners, lounges, an empty room early on |
| Warm-up | Steady groove, restrained bass, few big drops | Roughly the first hour of a night |
| Groovy | Bouncy bassline, busy percussion, swing | A floor that's starting to move |
| Peak | Full kick, heavy bass, big drops, dense mix | The busiest hour of the night |
| Dark | Gritty bass, sparse melody, tense, hypnotic | Late night, after the peak |
| Euphoric | Bright chords, big breakdowns, uplifting leads | Hands-in-the-air moments |
| Closing | Emotional, melodic, energy easing off | The last few tracks |
Four to seven labels is a good range. Fewer, and each folder becomes a grab bag. More, and neighboring labels start to sound alike, even to you.
If you'd rather think in numbers, energy levels work as labels too: Energy 1 to Energy 5, say. Keep the steps few. On a ten-point scale, the gap between a 6 and a 7 is too fuzzy for anyone, human or machine, to apply consistently.
How to classify music by mood automatically
Sorting a thousand tracks by mood by ear takes days. The faster way is to teach software your labels with examples, the way you'd brief a new resident: "these are warm-up tracks, these are peak tracks."
- Pick your axis and labels. Energy, mood, or a few combined labels, based on how you pick tracks mid-set.
- Make one folder per label. Fill each with tracks that clearly belong there. About 100 per label is a good target; you can start smaller and add more later.
- Train. The software listens to every example, measures rhythm, energy, and timbre, and learns what each label sounds like to you.
- Classify new music. Each new track gets the closest label plus a confidence score.
- Review the unsure ones. Low confidence usually means a track sits between two labels. Listen, decide, fix.
It's the same idea as classifying music into genre folders automatically, applied to feel instead of style. The software doesn't know what "dark" means in general. It learns what "dark" means in your folders, and that's the point: your Groovy isn't someone else's Groovy.
How to pick good example tracks for each label
The examples are the whole lesson, so they decide how good the results are.
- Pick the textbook tracks. Choose tracks you'd play for that moment without a second thought. If you hesitate, leave it out.
- Leave the in-betweens out. A track that's half Groovy, half Peak blurs the line between two labels. Classify it later; just don't teach with it.
- Mix artists, keys, and tempos inside a label. If every Dark example comes from the same producer, the software may learn that producer's sound instead of darkness.
- Keep labels roughly balanced. A label with 15 examples next to one with 200 will be recognized less reliably.
- One track, one folder. The same track in both Warm-up and Groovy teaches two opposite things at once.
- Judge the main part of the track. Not the intro or the breakdown, but the part that plays most of the time.
- Test yourself. Pull ten random tracks from a folder and listen. If you'd move two or more, tighten that label before expecting software to follow it.
FAQ
Can software really detect the mood of a song?
It can detect what mood sounds like: bright or dark tone, heavy or light bass, driving or relaxed rhythm. It learns that from your examples, so it follows your definition of each mood.
What it can't do is understand lyrics. A sad lyric over a happy beat sounds happy to it.
Can a track be both groovy and dark?
In real life, yes. For sorting, each track gets one label from your set. Choose the axis you rely on most when picking the next track, or create a combined label such as Groovy Dark if you use that pairing often.
How many mood labels should I start with?
Four to seven. Start with the ones you reach for at every gig, check how well the software tells them apart, then split or add labels later. Adding a label just means adding another example folder.
Should I use an energy scale from 1 to 10?
A shorter scale works better. Three to five energy levels, each clearly different from the next, classify far more consistently than ten levels where neighbors overlap.
Do I still need BPM and key?
Yes. Mood labels tell you which tracks fit the moment; BPM and key tell you which of those will mix smoothly. Filter by label first, then narrow by tempo and key. For shaping a whole night, see how to plan the energy flow of a DJ set.
Faster with Taxonomy DJ
Automatic music classification in Taxonomy DJ works with any labels you name, mood and energy included:
- In the Training tab, click Add label, name it, and choose a folder of example tracks, such as your Warm-up folder. All your mood folders inside one parent folder? Click From parent folder and each subfolder becomes a label. The app analyzes and trains on its own, then opens the Model report showing which labels get mixed up.
- In the Music Library tab, click Add music folder, choose Run full analysis in the After scanning box, then click Scan. Each track gets a label and a % Confidence.
- Click Quick filter ▸ Needs review to see only the unsure tracks, and fix labels in the Label cell or with the 1–9 keys.
- Filter the Label column, then narrow by BPM or Camelot in the filters above the column names to find tracks that fit the moment and mix well. Right-click ▸ Add to playlist to build your set list.
- Select tracks, then File ▸ Copy to folder: one subfolder per label, originals untouched.
Training and automatic labeling need Premium. BPM, key, labeling tracks by hand, and copying or moving files all work on the Free plan.
Read next: how AI music classification works and how to plan the energy flow of a DJ set.