EDT, T20 and T30: Why Reverberation Numbers Can Disagree

EDT describes the initial decay slope, while T20 and T30 estimate a sixty-decibel reverberation time from later fitting ranges. In conventional analysis, EDT uses 0 to −10 dB, T20 uses −5 to −25 dB and T30 uses −5 to −35 dB. Different slopes can produce different results from the same response.

The numbers are not three names for a stopwatch measurement from sound onset to silence. They summarize selected parts of a frequency-filtered decay curve. Understanding the fitted region is essential before comparing a measurement with a prediction or judging whether a room needs a particular treatment.

Separate a measured slope from a calculated prediction

The reverb calculator provides musical timing starting points from tempo. It does not record a room impulse response or calculate EDT from microphone audio. A room RT60 prediction instead uses entered room and absorption information under stated assumptions.

Measurement software starts with an actual response and analyzes its decay. Prediction starts with a model and inputs. Agreement can be informative, but one cannot verify the other merely because both display seconds. Retain the frequency band, positions, method and fitting range alongside every reported value.

Read the fitting windows

Conventional decay ranges and extrapolation
MeasureFitted dropIllustrative sixty-decibel extension
EDT0 to −10 dBSix times the fitted ten-decibel duration
T20−5 to −25 dBThree times the fitted twenty-decibel duration
T30−5 to −35 dBTwice the fitted thirty-decibel duration

EDT is commonly presented as a sixty-decibel-equivalent early-decay indicator. It is not the same later-field RT60 estimate as T20 or T30. The multipliers above describe an ideal straight-line illustration; software generally fits a regression line rather than timing only two selected points.

The RT60 formula guide explains the sixty-decibel concept. “T20” does not mean twenty seconds, and “T30” does not mean the sound disappeared after thirty decibels. The label identifies the fitting span.

An original numerical example

Imagine a filtered decay that reaches −10 dB after 0.12 seconds. Its early straight-line extrapolation gives EDT = 6 × 0.12 = 0.72 seconds. Suppose the same curve crosses −5 dB at 0.05 seconds, −25 dB at 0.45 seconds and −35 dB at 0.75 seconds.

The T20 interval is 0.45 − 0.05 = 0.40 seconds, giving 1.20 seconds after multiplication by three. The T30 interval is 0.75 − 0.05 = 0.70 seconds, giving 1.40 seconds after multiplication by two. All three values can describe different parts of this deliberately nonuniform decay.

These invented coordinates illustrate the arithmetic; they are not a measurement of a real room. A sharp initial drop followed by a slower late region can make an early indicator shorter than a late one. A single average would hide that behavior rather than explain it.

Why the noise floor and response length matter

A curve close to the background noise no longer provides clean evidence of the room’s decay slope. Extending a fit into that region can distort the estimate. T30 needs a larger usable decay range than T20, so an apparently more detailed label is not automatically a better result.

Likewise, a response cut off too soon can remove late energy. Check the retained impulse-response length, excitation and noise conditions before interpreting a suspiciously short number. A successful software calculation does not establish that the underlying data supported the method.

The prediction-versus-measurement guide explains additional differences involving absorption assumptions and spatial variation. Do not compensate for an unreliable fit by changing material coefficients until the displayed results happen to agree.

Use frequency and position context

In small rooms, low-frequency resonances often dominate behavior instead of a diffuse reverberant field. A neat RT60-style number at a bass frequency may therefore be less useful than a frequency-dependent decay view. The measurement’s label does not make the diffuse-field assumption true.

Use the room acoustics calculator for geometry-based planning and compare its limits with the room acoustics overview. Neither replaces position-specific measurements or proves the amplitude of an individual mode.

For repeatable comparisons, keep the same band definition, source and microphone positions, analysis settings and noise conditions. Then repeat at additional positions when the question concerns a whole room. One microphone location is not automatically representative of every seat or instrument position.

A result-review checklist

  1. Identify whether the value is predicted or measured.
  2. Record its band, fitting method and actual usable range.
  3. Inspect the decay curve rather than reading only the number.
  4. Check for truncation, noise contamination and low-frequency modal behavior.
  5. Compare like-for-like positions and settings before claiming an improvement.

A calculator workflow is useful for keeping model inputs explicit. Retain measured results separately so a convenient numerical agreement is not mistaken for independent validation.

Keep the original response alongside an exported table of results. A later reviewer can then inspect whether a changed number reflects a different fit, a different position or a genuine change in the measured response.

Frequently asked questions

Is T30 always more accurate than T20?

No. A longer fitting span can be useful only when the response has sufficient clean decay range. Noise, truncation or a changing slope can make the result unreliable or answer a different question.

Why can EDT be much shorter?

It follows the initial decay region, which can fall rapidly before a slower tail develops. Source direction and measurement position can affect that early behavior, so retain the actual curve and context.

Can I average all three values?

You can calculate an arithmetic mean, but it does not automatically become a meaningful room descriptor. Explain why the measures disagree before combining indicators that summarize different regions.

Does a short value prove good acoustics?

No. Purpose, frequency balance, reflections, modal behavior and spatial consistency also matter. A single decay time cannot certify a room’s quality or suitability for every use.

Can a hand clap supply a certified T30?

A casual clap can reveal audible problems, but it does not by itself establish a controlled source, calibrated method or sufficient decay range. Avoid attaching measurement certainty to an informal listening check.

Does this site measure EDT from a recording?

No such capability is claimed here. Its relevant room tools calculate predictions from user inputs. Use an appropriate measurement workflow when the question concerns an actual recorded impulse response.

Scroll to Top