FIELD GUIDE

Read every number honestly.

Each reading below states its unit, measurement window, practical interpretation and limit. SignalMetric favors transparent evidence over authoritative-looking guesses.

The critical boundary

dBFS and dBTP describe a digital signal relative to full scale. They are not calibrated dB SPL, dBA or a hearing-safety dose. Legal, occupational and certified delivery decisions require validated equipment and procedure.

0 dBFS ≠ 0 dB SPL

Digital level and headroomZero dBFS is the ceiling. Useful signals are negative values; the distance to zero is digital headroom.

Three loudness time scalesMomentary reacts in 400 ms, Short-Term describes 3 seconds, and Integrated accumulates a gated session.

Peak versus averageCrest, PSR and PLR compare peak energy with RMS, short-term loudness and integrated loudness respectively.

Spectral shapeCentroid is the energy-weighted center, bandwidth describes spread, and R85 marks the frequency below which 85% of energy falls.

Zero crossingsZCR counts sign changes. More high-frequency or noisy content usually crosses zero more often.

31 READINGS · 6 GROUPS

Digital full-scale. Not calibrated SPL.

Level & Peak

RMSdBFS

Window / method: Current analysis window; energy-averaged amplitude.
How to read it: Closer to 0 means a stronger digital input. It is usually steadier than a peak reading.
Limit: Not acoustic loudness or calibrated sound pressure.

Level & Peak

Sample PeakdBFS

Window / method: Highest stored sample in the current analysis window.
How to read it: Shows how close a captured sample came to digital full scale.
Limit: Can miss an inter-sample peak reconstructed between samples.

Level & Peak

Peak HolddBFS

Window / method: Highest recent sample peak with controlled visual decay.
How to read it: Useful for catching a short transient that the current value has already left behind.
Limit: A display memory, not a separate measurement standard.

Level & Peak

True Peak*dBTP

Window / method: 4× oversampled inter-sample estimate.
How to read it: Reveals reconstructed peaks that may exceed the highest stored sample.
Limit: Marked as an estimate; not a certified delivery-compliance result.

Level & Peak

HeadroomdB

Window / method: Distance from estimated True Peak to 0 dBFS.
How to read it: A larger positive value means more digital margin before full scale.
Limit: Does not describe analogue preamp or acoustic headroom.

Loudness

MomentaryLUFS

Window / method: BS.1770-style K-weighted mono window of 400 ms.
How to read it: Fast perceptual loudness evidence for phrases and short events.
Limit: Too reactive to represent an entire programme.

Loudness

Short-TermLUFS

Window / method: BS.1770-style K-weighted mono window of 3 seconds.
How to read it: Describes the current passage with less fluctuation than Momentary.
Limit: Still local in time; it is not the session average.

Loudness

IntegratedLUFS

Window / method: K-weighted, absolute- and relative-gated session accumulation.
How to read it: The best single loudness summary for the measured session.
Limit: Needs time to settle and does not certify a delivery master.

Loudness

LRALU

Window / method: Distribution of gated short-term loudness across the session.
How to read it: Higher values indicate greater programme loudness variation.
Limit: Shown as provisional before 60 seconds of useful material.

Loudness

Target ΔLU

Window / method: Integrated LUFS minus the selected visual reference.
How to read it: Positive is above the reference; negative is below it.
Limit: The Streaming, EBU and ATSC targets are guidance, not certification.

Dynamics

Crest FactordB

Window / method: Sample Peak minus RMS in the same analysis window.
How to read it: Larger values indicate sharper transients relative to average energy.
Limit: Sensitive to the current window and not the same as programme dynamics.

Dynamics

PSRdB

Window / method: Estimated True Peak minus Short-Term loudness.
How to read it: Describes short-passage peak-to-loudness contrast.
Limit: Depends on a stable Short-Term loudness reading.

Dynamics

PLRdB

Window / method: Estimated True Peak minus Integrated loudness.
How to read it: Summarizes session peak-to-loudness contrast.
Limit: Unstable early in a session while Integrated loudness is warming up.

Dynamics

Floor P10dBFS

Window / method: 10th percentile of RMS levels in the rolling 30-second history.
How to read it: A robust low-level baseline that ignores a few isolated silent frames.
Limit: An input-floor estimate, not a calibrated room-noise SPL.

Dynamics

SNR ESTdB

Window / method: Current RMS relative to the rolling Floor P10 estimate.
How to read it: Higher values suggest the current source stands farther above its recent baseline.
Limit: Not a laboratory SNR measurement and changes with the recent source.

Spectrum

DominantHz

Window / method: Strongest stable component in the current 2,048-point FFT.
How to read it: Often points to a fundamental tone, resonance or strongest partial.
Limit: Complex audio may not have one meaningful dominant frequency.

Spectrum

CentroidHz

Window / method: Power-weighted center of the current spectrum.
How to read it: Higher values generally correlate with a brighter spectral balance.
Limit: A shape descriptor, not the pitch or strongest frequency.

Spectrum

BandwidthHz

Window / method: Power-weighted spread around the spectral centroid.
How to read it: A wider value means energy is distributed across a broader frequency range.
Limit: Does not identify where separate spectral clusters occur.

Spectrum

R85Hz

Window / method: Frequency below which 85% of current spectral power falls.
How to read it: Compactly shows how far meaningful energy extends into the high range.
Limit: A percentile boundary, not a hard low-pass cutoff.

Spectrum

Flatness%

Window / method: Geometric-to-arithmetic mean ratio of spectral power.
How to read it: Low values are more tonal; high values are more noise-like.
Limit: Depends on the analyzed band and does not judge sound quality.

Spectrum

64-band FFTdBFS

Window / method: 2,048-point Hann FFT mapped to 64 logarithmic bands.
How to read it: Shows relative digital energy from roughly 45 Hz to 16 kHz.
Limit: Band values are not calibrated acoustic pressure or a laboratory RTA.

Spectrum

FFT ComponentsHz / dBFS

Window / method: Up to six strongest local FFT peaks with interpolation.
How to read it: Separates prominent partials by frequency, level, period, note and harmonic fit.
Limit: A selected peak view, not a complete inverse-FFT reconstruction.

Musical

Nearest Note

Window / method: Dominant stable frequency mapped to equal-tempered A4 = 440 Hz.
How to read it: Provides quick musical context such as A4 or F#3.
Limit: Approximate context, not a precision instrument tuner.

Musical

Cents Offsetcent

Window / method: Pitch distance from the nearest equal-tempered note.
How to read it: Positive is sharp and negative is flat relative to that note.
Limit: Only meaningful when the dominant frequency is stable and tonal.

Musical

Estimated BPMBPM

Window / method: Recent spectral-flux onset pattern within 48–200 BPM.
How to read it: Shows tempo only after repeated transient evidence becomes consistent.
Limit: Speech, sustained tones and free-time music may remain Listening.

Musical

Tempo Confidence%

Window / method: Consistency score for the recent beat estimate.
How to read it: Higher values mean the observed onset pattern supports the shown BPM more strongly.
Limit: Confidence is algorithmic evidence, not musical ground truth.

Signal Integrity

DC Offset EST%

Window / method: Mean sample value in the current time-domain window.
How to read it: Values near zero indicate the waveform is centered around digital zero.
Limit: A short-window estimate; route processing can influence it.

Signal Integrity

ZCR%

Window / method: Share of adjacent sample pairs that cross digital zero.
How to read it: Higher rates often accompany noisier or higher-frequency material.
Limit: Cannot identify content or quality by itself.

Signal Integrity

Clip Events

Window / method: Session count of entries into the digital clipping region.
How to read it: A rising count means the input repeatedly reached or approached full scale.
Limit: Does not prove audible distortion after route processing.

Signal Integrity

Channel CountCH

Window / method: Channel count reported by the active source format.
How to read it: Confirms whether the source is mono or multichannel before downmix analysis.
Limit: Multichannel analysis uses equal-gain averaging and can cancel opposite polarity.

Signal Integrity

Sample RatekHz

Window / method: Rate reported by the current microphone route or audio file.
How to read it: Shows how many samples per second feed the analysis.
Limit: A higher rate alone does not guarantee better source quality.