TrackRate user guide

Audio Metrics

Understand, calculate, display, sort by, and apply TrackRate's 21 audio measurements across a local music library.

7 questions

What are Audio Metrics and how do I calculate them?

Audio Metrics are 21 signal measurements that TrackRate computes by decoding each track — 18 raw metrics plus 3 composite derivative scores. They describe tempo, brightness, energy, dynamics, noisiness, and rhythmic density in numerical form, making it possible to filter, sort, and sequence your library by sound rather than only by metadata.

To calculate them, use Calculate Metrics from the toolbar or context menu. You can run it on selected tracks, a playlist, a folder, or the entire library. Analysis uses a worker thread so playback is not interrupted, and results are cached in the database so you only need to run it once per track. Batch calculation of a full library may take time — start with a playlist or folder to see results quickly.

Audio Metrics are a Premium feature. Without them, Metric Lens, Arrangers, metric track-row slots, and metric-based sorting are unavailable.

What does each metric actually measure?

The 21 metrics cover four broad dimensions of sound. Use this reference to pick the right measurement for what you are trying to find or shape.

Audio Metrics reference
DimensionMetricUnitWhat it tells you
Rhythm & PaceTempoBPMEstimated beats per minute
Tempo Confidence0–1How reliable the BPM estimate is
Beat Strength0–1Average beat magnitude — pulse-driven vs. fluid
Onset Rateonsets/sNote and beat attacks per second — busy vs. sparse
Zero Crossing Rate0–1Waveform sign-change rate — proxy for noisiness and high-frequency content
Brightness & TimbreSpectral CentroidHzBrightness — where the timbre centre of mass sits
Spectral RolloffHzFrequency below which 85% of energy lies
Spectral Flatness0–1Tonal (0) vs. noisy (1) — pure sine = 0, white noise = 1
Spectral Contrast0–1Difference between spectral peaks and valleys
Spectral FluxrawFrame-to-frame timbre change rate
Energy & LoudnessRMS PowerdBAverage loudness of the signal
Crest FactorratioPeak / RMS ratio — punchy and dynamic vs. flat and dense
Dynamic RangedBSpread between quietest and loudest moments
Dynamic Movement0–1How much loudness changes from moment to moment
Texture & WeightBass Ratio0–1Energy fraction below 250 Hz — bottom-heavy vs. bright
Sub-Bass Ratio0–1Energy below 60 Hz — physical weight and rumble
Mid-Range Density0–1Energy concentration in the vocal and instrument range
High-Freq Spread0–1Energy distribution above 8 kHz — air and sparkle
Derived ScoresEnergy0–1Composite perceived intensity combining loudness, density, and onset activity
Danceability0–1Composite of beat strength, tempo stability, and rhythmic regularity
Acousticness0–1Composite estimate of acoustic vs. electronic/synthetic origin
Which metrics should I pay attention to for my kind of music?

Start with the dimension that matches the listening decision you are trying to make, then narrow to one or two metrics within it. Using too many metrics at once produces results that are hard to interpret.

Practical starting points by goal

For DJ sets and beat-matching: Tempo, Beat Strength, and Onset Rate help find tracks with compatible BPM ranges and rhythmic density. For mood and energy shaping: Spectral Centroid (brightness), RMS Power (loudness), and the Energy derived score give you a reliable intensity gradient. For finding similar-sounding tracks: use Metric Neighborhood with Spectral Centroid, Spectral Flatness, and Bass Ratio — these three capture timbre, tonality, and low-end weight without overfitting. For dynamic contrast: Crest Factor and Dynamic Range separate punchy, breathing material from consistently dense, compressed tracks.

The three derived scores — Energy, Danceability, and Acousticness — are good first choices because they combine several raw measurements into musically intuitive dimensions. Switch to individual raw metrics when you need finer control over a specific characteristic.

How do I display metrics in track rows?

Each track row in the Library has two configurable metric slots positioned between the play count (♫) and the playlist star (☆). Open the Library header's settings or the sort menu to choose which metric occupies each slot. The selected metrics appear as compact numeric values on every visible track row.

Choose metrics that give you a quick read on the attribute you care about most. A common pairing is Tempo in slot one (for BPM at a glance) and Energy in slot two (for perceived intensity). When you change the slot assignments, all visible rows update immediately. Tracks that have not yet been analysed show blank slots — run Calculate Metrics to populate them.

Can I sort my library by audio metrics?

Yes. The Sort menu in the Library and Playlist headers includes every calculated metric as a sort option alongside the standard metadata and history fields. You can sort ascending or descending by Tempo, Spectral Centroid, Energy, Danceability, or any other metric that has been calculated for the visible tracks.

Metric sorting is most useful when combined with filters. For example, filter to #electronic with ratings 7–10, then sort by descending Energy to surface the most intense tracks first. Or filter to a single artist and sort by Tempo to see their slowest and fastest material side by side.

Tracks without calculated metrics sort to the bottom regardless of sort direction. Run Calculate Metrics on the filtered pool first if you need a complete ordering.

Why do some tracks show no metric values?

The most common reason is that Calculate Metrics has not been run on those tracks. Metrics are not computed automatically during scanning — you must trigger analysis explicitly. Other causes include:

Very short files (under a few seconds) may not contain enough signal for reliable measurement and are skipped. Corrupted or unreadable files that fail to decode will also show no values. Unsupported formats in edge cases may decode for playback but not produce analysable data through the worker-thread pipeline.

To check what is missing, filter for the tracks you expect to see, run Calculate Metrics on the selection, and watch the progress indicator. Any files that fail are logged. If a valid, playable track consistently fails analysis, it may use an unusual codec variant that the analysis decoder does not handle.

How do Arrangers use metrics differently from Metric Lens?

They use the same 21 measurements for fundamentally different purposes. Metric Lens is a filter — it reduces the eligible pool by keeping only tracks whose metric values fall inside chosen ranges. It is asking: "which tracks have these sonic properties?"

Arrangers use metrics to sequence a playlist — they order tracks so that metric values follow a deliberate shape across the listening session. An Energy Waves arranger might ramp energy up and down in arcs. A DJ Set arranger might smooth tempo transitions between adjacent tracks. Arrangers are asking: "in what order should these tracks play so the metrics flow well?"

Because they share the same data, Metric Lens and Arrangers compose naturally. Use Metric Lens to define a pool — say, #electronic tracks with BPM 120–130 and brightness in the mid-to-high range — then feed that filtered pool into an Arranger to sequence it. The lens narrows what is eligible; the arranger decides the order. See the Metric Lens and Arrangers guides for the full workflow.