Entering microphone response curve compensation in the equalization effect

Audacity Help Forum Windows
1

Initially I thought that a compensation curve (according to the microphone response curve which is in dB power ( SPL)) in the Equalization effect would have to be entered in dB Amplitude (half the values in the mic specs) in order to achieve a flat response on the low frequency end of a microphone recording because I saw the spectral plots and other explanations using the word “amplitude”. However, it seems that everything related to spectra and amplification is actually calculated in power (SPL) because I found the reference in the Help under “Amplify and Normalize” … “whose peak amplitude is -6 dB (+0.5 to -0.5 on the vertical scale)”. … so you can just use the positive values of dB found in the microphone specs for the equalizer curve to get a flat mic response? I guess the labeling as amplitude in the software is just something you have to know about.

Do you think that is correct?

Is there a way to get the spectral plots to display in dB Amplitude and/or linear Amplitude?

Thanks

2

Assuming the mic response is shown in dB then “decibels are decibels” and you compensate for a +6dB bump with a -6dB cut, etc.

+6dB is 2X the amplitude and 4X the power, but as long as you are working in dB everything is fine.

In the “electrical world” amplitude is voltage. So for example, 28V RMS into 8 Ohms is (about) 100 Watts. If you double the voltage you also double the current (Ohm’s Law) so you’ve got 400W and it’s 6dB louder.

And, if you double the digital amplitude you are doubling the voltage out of the DAC and doubling the voltage out of the amplifier (assuming no clipping).

3

There’s an app for that: Steve’s “SpectrumToEqCurve.ny”. It derives an equalization from a frequency-analysis.
Applying an inverted version of that equalization curve to the audio it was derived from &, voila flat-response.

4

Hi… and thanks for the reply…

well, this is my problem.

2 x amplitude is +3 dB. (10 log A1/A0) (or 0.5 amplitude is -3 dB)

2 x power (SPL) is +6 dB (20 log A1/A0) (or 0.5 SPL is -6 dB)

Therein is my question why it’s called amplitude, but seems to be the mathematics of power. Or we just call it amplitude in the software docs anyway, and just remember that it is non-standard?

Cheers

5

OK… I just did a little physics review and see where my confusion lies.

Audacity is using and labelling the amplitude of the measured voltages (from a microphone or whatever) which has the same
dB calculation as sound pressure level (SPL)

Whereas I am thinking in pure physics and relating the equations for dB for amplitudes of the actual sound intensities (waves). (like 120 dB being the threshold of pain etc.)

Thanks a lot for making me think about it more.

6

We came in in the middle of the movie. Why do we want to do this? It may seem that having a ruler-flat microphone is a good thing but it’s a safe bet that it’s not the case.

Behold Nataly Dawn from Pomplamoose.

NatalyAndU87.png
She’s standing in front of a U87 microphone. Anybody who has ever heard this combination agrees that the microphone loves Nataly’s voice. It’s not flat. It has some interesting lumps and bumps in the response that happen to get along well with voices, particularly hers.

You can correct some intentional distortions such as The Shure SM58 rock band microphone “presence boost.”

SM58PresenceBoost.png
And much more of a current curse, the Sibilance Boost that many home microphones are saddled with.

SibilanceBump.png
Nobody would complain about correcting either one of those.

This is a Shure A15HP rumble filter which is used in line with your broadcast field microphone to keep traffic and wind noise out. You can just make out the characteristic printed on the side. The bend is about 100Hz.

A15HP.png
A version of this is included with the Audacity Audiobook Mastering Suite of tools. Oddly, one reason that filter is included with Audiobok Mastering is a very common error that many home microphones have. They have very low frequency data errors that nobody bothers to correct. Why bother? Who is going to know?

If you make one of these “flat,” you may destroy the performance.

So what’s the goal?

If you’re doing education and training exercises, it’s good to know that Audacity does occasionally takes shortcuts in order to sound better. This can drive scientists nuts because the math doesn’t always work out. But the sound usually does.

Koz

7

2 x amplitude is +3 dB. (10 log A1/A0) (or 0.5 amplitude is -3 dB)

2 x power (SPL) is +6 dB (20 log A1/A0) (or 0.5 SPL is -6 dB)

No…

The power formula is dB = (10 log P/Pref)
The amplitude formula is dB = (20 log A/Aref)

A +3dB increase in amplitude (~1.4X) is a +3dB increase in power to the speakers (2X) and a +3dB increase in SPL level.

As above, if you have 100W and you double the amplitude (voltage) you get 4 times the power (400W). But no matter how you cut it, that’s a 6dB increase.

BTW - Digital levels are dBFS (decibels full scale) where 0dB is the “digital maximum”. It’s as high as you can “count” with 16 bits*, etc. That’s why digital levels are usually negative.

The dB SPL reference is approximately the quietest sound that can be heard so SPL levels are normally positive.

Normally, nothing is calibrated but everything is directly correlated so if your digital level goes-down by 6dB the SPL level will also go down by 6dB.

Therein is my question why it’s called amplitude, but seems to be the mathematics of power.

By default, the waveforms in Audacity show the normalized amplitude. You’re seeing a graphical representation of the actual audio sample values. ( If you don’t understand sample values, see [u]Digital Audio Fundamentals[/u].) That is, if you load a 16-bit file that peaks at 0dB, the “raw numbers” go from −32,768 to 32,767**. Those numbers are normalized and converted to floating-point so they go from -1.0 to +1.0 and that’s what you see on the waveform scale. A file peaking at -6dB will have amplitude values half that (shown as -0.5 to +0.5).

There is no power in the digital file. Just a series of numbers, each representing the amplitude of the wave at one instant in time.






\

  • Everything is automatically scaled by the drivers so a 0dB 8-bit 0dB file is just as loud as a 0dB 24-bit file. (In floating-point 0dB is calibrated to 1.0.)

** It’s complicated but since there is no “negative zero” so you get one extra “bonus count”.

8

Thanks… I think I understand how the software is working now. I am thinking “amplitude of the sound waves” and it works using “amplitude of the voltage or SPL” which is already squared. (power of the sound wave wiggle traces if you like)

see:

http://hyperphysics.phy-astr.gsu.edu/hbase/Sound/intens.html#c4

Cheers

9

Sorry… what I was thinking was that the amplitude was the “amplitude of the pressure waves”, so that’s where I got confused.

Cheers