The Tone-to-Noise Ratio quantifies the energy of a discrete tonal component relative to the
broadband noise energy within the same critical band. TNR is widely used in product noise
certification, IT equipment standards, and environmental noise assessment to objectively
determine whether tonal content is prominent enough to warrant a penalty or corrective action.
Supported Standards
Standard
Scope
Edition
ECMA-418-1 Annex A
Psychoacoustic tonality based on Aures/Terhardt model
Edition 3 (2022)
ECMA-74
Measurement of airborne noise emitted by information technology and telecommunications equipment
Edition 18 (2023)
ECMA-74 defines the TNR measurement procedure for IT equipment (servers, workstations,
printers, etc.) and references ECMA-418-1 Annex A for the detailed computation algorithm.
The method is also applicable to general product noise evaluation beyond the IT domain.
Theory
Critical-Band Analysis
The input signal is analyzed using a high-resolution FFT. For each detected tonal component,
the corresponding critical band is identified. The energy within that critical band is
partitioned into tonal energy (discrete peak) and broadband noise energy (remaining spectral
content):
TNR = 10 · log10(Wtone / Wnoise) [dB]
Here Wtone is the power of the tonal peak (summed over a narrow bandwidth around
the peak) and Wnoise is the total noise power within the critical band after
subtracting the tonal contribution.
Tone Identification
Tonal peaks are identified through a spectral peak-picking algorithm. A candidate peak must
satisfy the following criteria:
The peak level exceeds the local noise floor by at least a minimum threshold
The peak bandwidth is narrow (typically less than the critical bandwidth at that frequency)
Adjacent spectral lines on both sides of the peak show decreasing levels
Prominence Assessment
ECMA-74 classifies tones based on their TNR value using frequency-dependent thresholds:
Prominent tone: TNR ≥ ΔTPR(f)
Not prominent: TNR < ΔTPR(f)
The threshold ΔTPR(f) varies with frequency, reflecting the ear's
frequency-dependent sensitivity to tonal content in noise. Typical threshold values range
from 6 dB at low frequencies to 8 dB at frequencies above 1 kHz.
Parameters
Parameter
Options
Description
FFT Size
8192 – 65536 samples
Spectral resolution; larger size improves tone separation from noise floor
Window Function
Hanning (default) / Flat-top / Rectangular
Windowing applied before FFT; Hanning is specified by ECMA-74
Number of Averages
1 – 200
Number of spectral blocks averaged; more averages improve noise floor stability
Frequency Range
50 Hz – 16000 Hz
Analysis bandwidth for tone search
Signal Type
Stationary / Time-varying
Selects single-result or running analysis mode
Output Quantities
Quantity
Unit
Description
TNR
dB
Tone-to-noise ratio for each detected tone
Tone Frequency
Hz
Center frequency of the detected tonal component
Tone Level
dB SPL
Sound pressure level of the tonal component
Noise Level
dB SPL
Broadband noise level within the critical band
Prominence Flag
Yes / No
Whether the tone exceeds the prominence threshold
TNR(t)
dB
TNR vs Time trace for time-varying signals
TNR vs Time
When operating in time-varying mode, the TNR is computed for each analysis block independently.
The resulting TNR(t) trace is used to:
Assess temporal stability of tonal prominence during product operation cycles
Identify transient tonal events that may be missed in averaged stationary analysis
Correlate tonal emergence with operating conditions (e.g., fan speed changes, compressor
cycling)
Interpretation Guidelines
ECMA-74 classification: A tone is classified as "prominent" when the TNR
exceeds the frequency-dependent threshold. Prominent tones trigger reporting requirements
and may necessitate product redesign to meet noise emission declarations.
TNR < 0 dB — Tone energy is below broadband noise energy in the critical
band; tone is masked and inaudible.
0 – 6 dB — Tone is present but not prominent; typically does not warrant
a penalty.
6 – 10 dB — Tone is at or near the prominence threshold; may be classified
as prominent depending on frequency.
> 10 dB — Clearly prominent tone; high annoyance potential and likely to
trigger a tonal penalty.
Practical Application Notes
ECMA-74 specifies a Hanning window with specific FFT size requirements based on sample
rate. Always use the standard-prescribed parameters when testing for compliance.
For IT equipment certification, TNR must be reported at each declared operating mode
(idle, active, maximum). The worst-case TNR across modes determines the tonal
classification.
TNR and tonality (DIN 45681) assess similar phenomena but use different mathematical
frameworks. TNR operates in the linear energy domain, while DIN 45681 uses a
masking-based audibility model. Results may differ for complex multi-tone signals.
When broadband noise levels are very low, TNR values can be artificially inflated.
Ensure that the signal-to-noise ratio of the measurement chain is sufficient (background
noise at least 10 dB below the device noise).
The critical-band partitioning is essential: a tone near the boundary of two critical
bands may be split, yielding a lower TNR than expected. Use sufficient frequency resolution
to accurately capture the tonal peak.
Standard References
ECMA-418-1:2022 — Psychoacoustic metrics for ITT equipment — Part 1: Prominent
discrete tones (Edition 3), Annex A
ECMA-74:2023 — Measurement of airborne noise emitted by information technology and
telecommunications equipment (Edition 18)
ISO 7779:2018 — Acoustics — Measurement of airborne noise emitted by information
technology and telecommunications equipment
E. Terhardt, G. Stoll & M. Seewann, "Algorithm for extraction of pitch and pitch salience
from complex tonal signals," JASA, vol. 71, 1982