Mixes that sound unclear, muddy, harsh, or dull often have hidden frequency-domain causes that are invisible on a standard level meter. Five-band energy analysis reveals which frequency regions dominate and which are recessed, making tonal balance problems visible and actionable without requiring you to interpret a full spectral chart.
What five-band energy analysis reveals (and why standard metering misses it)
Your mix can pass every loudness standard, sit at the correct RMS level, and still sound wrong. The problem lives in the frequency domain: too much low-mid energy creates muddiness, excessive presence creates harshness, and recessed highs make the mix sound veiled. Standard level meters measure overall amplitude but cannot tell you which frequency regions are responsible for these perceptual problems.
Five-band energy analysis is a spectral measurement technique that divides the frequency spectrum into five perceptually meaningful regions to assess tonal balance (Source: inputs/articles/5-band-energy/brief.md#Core message). Instead of showing you a full spectral chart with hundreds of frequency bins, it collapses the spectrum into five bands that correspond to how we hear and describe sound: Sub, Low-Mid, Mid, Presence, and High.
This compression makes tonal balance problems immediately visible. When the Low-Mid band exceeds the Mid band by 5 dB or more, you see potential muddiness at a glance. When the Presence band exceeds the Mid band by 8 dB or more, you see harshness risk. These relationships are difficult to extract from a full spectral chart but instant in a five-band view.
The five-band approach also feeds automated diagnostic systems. These measurements serve as the primary input for tonal balance summary generation, allowing AI coaching engines to identify specific frequency problems and generate corrective recommendations without manual spectrum inspection.
The five frequency bands and what they contain
The five bands divide the 20 Hz to 20 kHz spectrum into regions that align with how we perceive and describe tonal character in music production.
Sub (20–100 Hz) contains sub-bass, kick drum fundamental, and bass extension (Source: inputs/articles/5-band-energy/brief.md#Page structure - The five frequency bands explained). This is the region you feel more than hear on large monitoring systems. When elevated, this band produces rumble or excessive sub-bass presence. When recessed, the mix lacks weight and low-end extension.
Low-Mid (100–500 Hz) contains bass body, kick and bass fundamentals, and the muddiness zone (Source: inputs/articles/5-band-energy/brief.md#Page structure - The five frequency bands explained). Most of the energy in bass-heavy genres lives here. This is also where congestion and masking occur when too many low-frequency instruments compete. Boxiness and muddiness are perceptual symptoms of excessive Low-Mid energy relative to the Mid band.
Mid (500 Hz–4 kHz) contains vocals, guitars, keys, snare, and the core of most arrangements (Source: inputs/articles/5-band-energy/brief.md#Page structure - The five frequency bands explained). This is the region where human hearing is most sensitive. A weak midrange makes the mix sound hollow or distant. An excessively strong midrange can produce nasal or boxy tonality, depending on which part of the band dominates.
Presence (4–8 kHz) contains vocal clarity, sibilance, attack definition, and the peak of human ear sensitivity (Source: inputs/articles/5-band-energy/brief.md#Page structure - The five frequency bands explained). This band controls intelligibility and perceived aggression. Too much presence energy creates harshness and listening fatigue. Too little presence makes the mix sound dull or muffled, even if the High band contains adequate energy.
High (8–20 kHz) contains air, shimmer, cymbals, and digital brightness (Source: inputs/articles/5-band-energy/brief.md#Page structure - The five frequency bands explained). This region determines whether a mix sounds open and airy or dark and closed. Excessive high energy produces harshness distinct from presence-band harshness—it manifests as sizzle rather than bite. Rolled-off highs produce a veiled sound that lacks sparkle.
| Band | Frequency Range | Musical Content | Common Problems |
|---|---|---|---|
| Sub | 20–100 Hz | Sub-bass, kick fundamental | Excessive rumble, lack of sub presence |
| Low-Mid | 100–500 Hz | Bass body, muddiness zone | Congestion, masking, boxiness |
| Mid | 500 Hz–4 kHz | Vocals, guitars, snare | Weak midrange, nasal tonality |
| Presence | 4–8 kHz | Vocal clarity, sibilance | Harshness, listening fatigue |
| High | 8–20 kHz | Air, shimmer, cymbals | Excessive sizzle, lack of air |
How the measurement is performed
Five-band energy analysis derives from Short-Time Fourier Transform (STFT) spectral analysis. The measurement uses STFT with a 2048-sample window, 512-sample hop, and Hann window (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). These parameters balance frequency resolution with time-domain averaging, producing stable measurements that reflect the overall spectral character of the mix rather than instantaneous variations.
How is 5-band energy measured? Using STFT with a 2048-sample window, 512-sample hop, Hann window, on mono downmix, averaging magnitude spectrum across time frames.
The analysis is performed on a mono downmix for consistency (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). This eliminates stereo-width variations that could skew energy measurements in individual bands. The mono downmix ensures that left-right balance does not affect the tonal balance assessment, which is the measurement's primary purpose.
Once the STFT produces a magnitude spectrum for each time frame, the magnitude spectrum is averaged across all time frames and converted to decibels (Source: inputs/articles/5-band-energy/brief.md#Page structure - How the measurement is performed). The dB conversion produces values that correspond to perceived loudness relationships rather than linear amplitude, which aligns the measurement with how we hear spectral balance.
After converting to dB, the mean value is extracted for each of the five frequency bands. The resulting values are signed: higher (less negative) values indicate more energy in that band. A band with -15 dB contains more energy than a band with -20 dB.
How to read and interpret the values
The values produced by five-band energy analysis are relative spectral amplitudes, not absolute loudness measurements (Source: inputs/articles/5-band-energy/brief.md#Page structure - How to read and interpret the values). This distinction is critical. A Sub band reading of -12 dB does not mean the sub-bass is quiet in an absolute sense. It means the sub-bass energy, averaged across the entire spectrum, sits at -12 dB relative to the peak magnitude in the spectrum.
What matters is the relationship between adjacent bands, not the absolute dB values. A well-balanced mix might show Sub at -10 dB, Low-Mid at -12 dB, Mid at -15 dB, Presence at -18 dB, and High at -22 dB. These absolute values tell you little on their own. The relationships between them reveal the tonal character.
When the Sub band exceeds the Low-Mid band by 8 dB or more, you have sub-bass buildup that needs attention (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). This indicates excessive low-frequency energy that may cause the mix to sound boomy or ill-defined on systems with extended bass response.
When the Low-Mid band exceeds the Mid band by 5 dB or more, you have potential muddiness or congestion (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). This is the single most common tonal balance problem in home studio mixes. Too much energy in the 100–500 Hz region masks midrange detail and creates a cloudy, indistinct sound.
When the Presence band exceeds the Mid band by 8 dB or more, you have harshness risk (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). This produces listening fatigue and makes vocals sound aggressive or sibilant. It is the opposite problem from muddiness—too much energy in the clarity and attack region.
When the High band sits 10 dB or more below the Presence band, you have veiled or dark sound (Source: inputs/articles/5-band-energy/brief.md#Key accuracy requirements). The mix lacks air and sparkle. This is common in mixes where high-frequency content has been overly attenuated to avoid harshness, creating a dull tonality instead.
| Comparison | Threshold | Indicates |
|---|---|---|
| Sub > Low-Mid | 8+ dB | Sub-bass buildup |
| Low-Mid > Mid | 5+ dB | Muddiness/congestion |
| Presence > Mid | 8+ dB | Harshness risk |
| High < Presence | 10+ dB | Veiled/dark sound |
Common scenarios and what the bands reveal
Five-band energy patterns reveal genre-specific and problem-specific tonal signatures.
A bass-heavy electronic mix typically shows an elevated Sub band, with Sub exceeding Low-Mid by 6–10 dB. Low-Mid, Mid, and Presence bands follow a gradual rolloff, and the High band is moderately recessed. This is intentional genre convention, not a problem. The Sub elevation creates the weight and power expected in electronic music. The key diagnostic question is whether the Sub-to-Low-Mid relationship exceeds 10 dB, at which point the mix may sound unbalanced even within genre expectations.
A muddy rock mix shows Low-Mid exceeding Mid by 6 dB or more. The Sub band is moderately elevated, Low-Mid is dominant, and Mid is recessed relative to both. This creates the characteristic cloudy, indistinct sound where guitars, bass, and kick occupy the same frequency space without separation. The Presence and High bands may be elevated in an attempt to create clarity, but this produces harshness rather than solving the underlying congestion.
A bright pop mix shows elevated Presence and High bands. Mid sits at a moderate level, Low-Mid is recessed relative to Mid, and Sub is further recessed. This produces a clear, forward, aggressive sound. The Presence band often exceeds the Mid band by 5–7 dB, approaching but not exceeding the 8 dB harshness threshold. The High band sits only 5–8 dB below Presence, creating air and sparkle without darkness.
A dark acoustic mix shows a significantly recessed High band. Sub and Low-Mid are moderate, Mid is slightly elevated, Presence is moderate or slightly recessed, and High sits 12–15 dB below Presence. This produces a warm, closed tonality. The mix may sound pleasant on small speakers but lacks the air and extension that modern streaming platforms favour.
A balanced commercial mix shows a gradual rolloff from Sub to High. Sub is the highest, Low-Mid is 2–4 dB lower, Mid is 3–5 dB lower than Low-Mid, Presence is 3–5 dB lower than Mid, and High is 8–10 dB lower than Presence. This creates a neutral tonal profile that translates well across playback systems. No single band dominates, and the thresholds for muddiness, harshness, and veiled sound are not exceeded.
Five-band energy analysis reveals tonal balance problems by comparing relative energy between adjacent frequency bands: sub-bass buildup when Sub exceeds Low-Mid by 8 dB, muddiness when Low-Mid exceeds Mid by 5 dB, harshness when Presence exceeds Mid by 8 dB, and veiled sound when High is 10 dB below Presence.
How 5-band energy feeds into automated systems
Five-band energy measurements do not exist in isolation. They serve as the primary input for tonal balance summary generation (Source: inputs/articles/5-band-energy/brief.md#Page structure - How 5-band energy feeds into other outputs). The tonal balance summary reads the five-band values, applies the diagnostic thresholds described above, and produces a plain-language assessment: "Low-Mid energy exceeds Mid by 6.2 dB, indicating potential muddiness."
These values are passed verbatim to the AI coaching engine that generates specific EQ recommendations (Source: inputs/articles/5-band-energy/brief.md#Page structure - How 5-band energy feeds into other outputs). The coaching engine does not re-analyse the audio. It reads the five-band values, identifies which thresholds are exceeded, and generates corrective guidance based on the specific relationships between bands. If Low-Mid exceeds Mid by 6 dB, the engine recommends attenuation in the 200–400 Hz region. If Presence exceeds Mid by 9 dB, it recommends de-essing or high-mid attenuation.
The five-band analysis is also combined with spectral centroid for brightness context (Source: inputs/articles/5-band-energy/brief.md#Page structure - How 5-band energy feeds into other outputs). Spectral centroid measures the centre of mass of the frequency spectrum. A mix can have elevated Presence energy (which five-band analysis detects) but a low spectral centroid (indicating that most energy still sits in the low-mid region). This combination reveals that the harshness is localised to specific transients rather than being a consistent characteristic of the mix.
Automated diagnostic systems use five-band energy as the first-pass filter. Full spectral data is computationally expensive to process and difficult to summarise in natural language. Five-band analysis produces a small, interpretable data structure (five numbers) that reveals most tonal balance problems without requiring manual inspection of a 1024-bin magnitude spectrum.
When to use 5-band versus full frequency response curves
Five-band energy analysis and full frequency response curves derive from the same STFT data, but they serve different diagnostic purposes.
Five-band analysis provides quick high-level assessment and problem identification. Use it when you need to answer the question: "Does this mix have a tonal balance problem, and if so, where?" The five-band view makes muddiness, harshness, sub-bass buildup, and veiled sound immediately visible. It is the diagnostic tool you use first, before detailed inspection.
Full frequency response curves provide detailed analysis, reveal narrow resonances, and identify precise EQ targets (Source: inputs/articles/5-band-energy/brief.md#Page structure - Relationship to full frequency response curves). Use the full frequency response curve when you already know a problem exists and need to locate the specific frequency. A five-band analysis might reveal Low-Mid exceeding Mid by 6 dB, indicating muddiness. The full curve shows you that the problem is a 3 dB peak at 280 Hz, not a broad buildup across the entire Low-Mid band.
Both measurements are derived from the same STFT data. The five-band measurement averages all frequency bins within each band. The full curve preserves individual bin data. Neither is more accurate—they serve different levels of diagnostic detail.
In automated systems, five-band energy feeds the tonal balance summary and automated problem detection. The full curve is available for manual inspection when the engineer wants to see detailed spectral behaviour. The five-band view answers "what is wrong", and the full curve answers "exactly where".
Summary and key takeaways
Five-band energy analysis divides the 20 Hz to 20 kHz spectrum into five perceptually meaningful regions: Sub (20–100 Hz), Low-Mid (100–500 Hz), Mid (500 Hz–4 kHz), Presence (4–8 kHz), and High (8–20 kHz). Each band corresponds to distinct musical content and common tonal problems.
The diagnostic power of five-band analysis comes from comparing relative energy levels between adjacent bands, not from interpreting absolute dB values. Sub exceeding Low-Mid by 8 dB or more indicates sub-bass buildup. Low-Mid exceeding Mid by 5 dB or more indicates muddiness. Presence exceeding Mid by 8 dB or more indicates harshness. High sitting 10 dB or more below Presence indicates veiled or dark sound.
These measurements feed automated tonal balance assessment and AI coaching systems. The five-band structure compresses complex spectral data into five interpretable values, making frequency-domain problems visible without requiring engineers to analyse full spectral charts. This compression enables automated natural-language summaries and corrective recommendations.
Use five-band analysis for first-pass tonal balance assessment. Use full frequency response curves when you need to locate specific resonances or narrow-band problems. Both derive from the same spectral data and serve complementary diagnostic roles.