I just noticed, perhaps naively, that the representation of amplitude on the y-axis of a waveform is somewhat paradoxical. Although the space between each value on the y-axis is identical, the units of amplitude are a logarithmic representation of the intensity of the sound.

I'm no acoustician, so maybe this is just "how it's done", but doesn't this influence perhaps in ways we just take for granted how we analyze a waveform?

  • Can you give an example of a program which scales values on the y-axis of a waveform logarithmically? Praat does not by default, for instance. Or are you speaking of the product of analyzing the raw waveform, such as RMS amplitude values?
    – user6726
    Feb 14 '16 at 16:46
  • Yes, I apologize because I think I've misunderstood your question. Are you talking about Praat's Intensity contour, or the waveform? Feb 15 '16 at 18:04

It’s not practical or reasonable to scale sound intensity linearly. It’s impractical because the resulting visuals would be unreadable, and it’s unreasonable because (simplifying a little) humans perceive loudness on a logarithmic scale. To crib a common example, imagine you’re listening to a single violin. To double that loudness, you’d need 10 violins, not 2.


In recording and audio editing, waveforms are always (in my experience) represented linearly. It is simply time vs. amplitude, both linear. In meters, however, i.e. the green/yellow/red lights that jump up and down on stereos, it is usually represented logarithmically, in order to fit it into the range of a few lights. In what context have you seen waveforms represented logarithmically?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.