Audio Software
Audio software is full of decibel readings, and almost none of them is the quantity noise limits are written in. Reading one as the other is a specific and common error.
When acoustic measurement is part of a wider workplace review, the linked guide can add operational context, but it should not be treated as a substitute for calibrated sound data.
The units are different
A sound level meter reports dB SPL — sound pressure in air, referenced to 20 micropascals. That is the quantity in every exposure limit.
Audio software reports dBFS — decibels relative to full scale, where 0 is the loudest value the digital format can represent and everything else is negative.
For additional technical context, see Svantek's sound measurement guide.
There is no fixed relationship between them. A recording peaking at −6 dBFS could have been made at 60 dB SPL or 110, depending on the microphone, the preamp gain and the converter.
The conversion requires calibration, which is the same requirement as anywhere else and is the thing that turns a recording into a measurement.
What each kind of tool is for
Editors and recorders. Capturing a signal. Their meters exist to prevent clipping, not to describe the room.
Spectrum analysers. Showing energy across frequency, which is genuinely useful for diagnosing what a noise consists of — a low-frequency hum and a high-frequency whine call for different remedies.
Loudness meters — LUFS and similar. Built for broadcast and streaming normalisation, weighted for perceived loudness of programme material. A different quantity again, and not an occupational one.
And room analysis tools, which measure reverberation and response using a known test signal. These answer the question a level meter cannot about why a room is difficult.
Where software genuinely helps
Identifying the source. A spectrum showing a strong peak at a specific frequency points at a specific machine, and that is what a dosimeter cannot tell you.
Reverberation time, which decides speech intelligibility and is not visible in a level reading.
And documenting a pattern over time, where a long recording shows when a noise occurs even if the absolute levels are unanchored.
The calibration route
If you want sound pressure numbers from software, the chain must be calibrated end to end.
A calibrator applied to the microphone, with the software's reading noted at that known level, establishes the offset.
Everything downstream must then stay fixed — gain, input settings, processing. Any change invalidates it, which is why this is done for a session rather than once.
And the result is still not a compliance measurement unless the whole chain meets an instrument specification, which is what a meter class describes.
The short version
- Sound level meters report dB SPL, referenced to air pressure; audio software reports dBFS, referenced to digital full scale
- There is no fixed relationship between them — a peak at −6 dBFS could be 60 or 110 dB SPL depending on the chain
- Editor meters exist to prevent clipping, loudness meters serve broadcast normalisation, and neither is an occupational quantity
- Spectrum analysis genuinely helps by identifying which source a noise came from, which dosimetry cannot
- Room analysis answers why a space is difficult, which a level reading does not
- Calibrating the whole chain gives sound pressure numbers for that session only, and still is not a compliance measurement