As usual, I decided get myself involved in something and wound up chasing my tail. Although my career is in engineering, and I have a past history with audio: as far as video goes, I'm just a hobbyist.

I decided to google up what a proper reference level would be for a 1khz test tone. It seems it all depends on which standards body you pay homage to, but I did manage to find a reference pointing to SMPTE. They shoot for -20dbFS (peak meter) to equal 0dbu (VU meter). Sounds reasonable to me - so I jumped into SoundForge 9 and generated a test tone for Vegas.

The test tone read -20dbFS, -5dbvu. Is this normal? I would think it would be zero, but then again there's that religion issue with standards.

Mean while, going back to SoundForge, I had some oddball value on the VU meter there too. Luckily, SoundForge lets you set the reference level - there it was - set 0dbu to -20dbFS for SMPTE. Let's see - now we're reading -3dbu. Close - but not 0. On a hunch, I fed it a -20dbFS square wave, and sure enough, the VU read 0dbu. It looks like SoundForge would rather have you down -3dbvu for an RMS sine in case your a wild guy playing with square and other odd shaped waves.

So why not play that -20dbFS square thru Vegas? Not what I expected - now it reads as -18.2dbFS and the VU goes -2.1dbu. Dammit - now I'm really hooked and I've got to get to the bottom of it.


I decided to create some fullscale waves in SoundForge (16bit 192khz). Started with a 0dbFS sine, and it read 0dbFS. Made a square at 0dbFS - it too read 0dbFS. Of course, the VU's were off the chart. Just for grins I go in to look at individual sample. well now: -3,276, and +3276 . (I suppose that SoundForge is just doing some scaling in the dialog - we all know the +32767 and -32767 are the min and max samples in a 16bit wave). But this is interesting. Going back to -20dbFS references- peak meter in SoundForge reads the same whether the wave is square or sine, but the peak sample values of both are the same.

It looks like the peak meter in SoundForge is some sort of absolute peak meter with a long decay - it isn't RMS. However, the VU in Soundforge does appear to be RMS - it reads 3db higher when playing a square wave (of the same peak value as a sine). I don't know whether this is good, bad, or indifferent for SoundForge - but damn if did'nt point out some big differences in Vegas.

In Vegas, a -20dbFS sine reads as -5dbu on the VU. (We'll buy the -5dbu for now - it's on the decibel scale, right?.) But a -20dbFS square (absolute peak same as sine - from SoundForge) read as -18.2dbFS and
-2.1dbu - somewhat odd - but some observations.

Not sure what the peak meter is in Vegas - it doesn't read absolute peaks as in SoundForge, nor is it some sort of an RMS value, as far as I can tell since it's difference is 2.2. (I have the suspicion that this is some sort of a magic number that might give someone a clue as to how the peak meter in Vegas reads) The VU meter in Vegas though, is off by 2.9 - almost close enough for goverment work to conclude it's doing some sort of RMS average.

So - Vegas reads different peak values for sines and squares with identical peak sample values. I'm not sure what to think of it....is this normal?

Also, I wonder if there is some way the set the VU 0dbu reference in Vegas like it can in SoundForge....

Anyway - you can always say for sure that the VU will always read -5dbu in Vegas (at least on my installation), based on sine at -20dbfs peak...

Hi Gene,

Wow, how did you come up with this can of worms?
Let me see if I can shine some light on your dilemma or just confuse you even more;)
There are so many standards, school of thought and myths that I’m sure someone will have different opinions, but here is my explanation:


decided to google up what a proper reference level would be for a 1khz test tone. It seems it all depends on which standards body you pay homage to, but I did manage to find a reference pointing to SMPTE. They shoot for -20dbFS (peak meter) to equal 0dbu (VU meter). Sounds reasonable to me - so I jumped into SoundForge 9 and generated a test tone for Vegas.

The test tone read -20dbFS, -5dbvu. Is this normal? I would think it would be zero, but then again there's that religion issue with standards.

Mean while, going back to SoundForge, I had some oddball value on the VU meter there too. Luckily, SoundForge lets you set the reference level - there it was - set 0dbu to -20dbFS for SMPTE. Let's see - now we're reading -3dbu. Close - but not 0.


If as you say -20dBFS is equal 0dBu and on your VU meter a sine tone reads -5dBVU/-3dBVU, then I would say that it sounds about right, but not quiet...
The reason is because 0VU is actually +4dBu not 0dBu. So you are missing a dB in your reading.
Did you round it up/down or something?


On a hunch, I fed it a -20dbFS square wave, and sure enough, the VU read 0dbu. It looks like SoundForge would rather have you down -3dbvu for an RMS sine in case your a wild guy playing with square and other odd shaped waves.

So why not play that -20dbFS square thru Vegas? Not what I expected - now it reads as -18.2dbFS and the VU goes -2.1dbu. Dammit - now I'm really hooked and I've got to get to the bottom of it.


A VU meter is just a RMS volt meter with a very specific scale (not a peak meter). As you now know, it is calibrated to indicate 0VU when 1KHz sine wave is applied to it at +4dBu. A sine wave has a different RMS voltage then a square wave has at the same amplitude. To spare you the google search, here are the formulas:

For sine wave Vrms = Vpk / sqrt(2)
For square wave Vrms = Vpk

So let say we have two signals (sine and square) with identical amplitude of 1Vpeak. The RMS voltage for sine wave will be 0.707 while the square wave RMS voltage is 1. Applying dB calculations to these two values, you will find that they are 3dB apart [20*log (0.707 / 1) = -3dB]. Exactly what you found.



I decided to create some fullscale waves in SoundForge (16bit 192khz). Started with a 0dbFS sine, and it read 0dbFS. Made a square at 0dbFS - it too read 0dbFS. Of course, the VU's were off the chart. Just for grins I go in to look at individual sample. well now: -3,276, and +3276 . (I suppose that SoundForge is just doing some scaling in the dialog - we all know the +32767 and -32767 are the min and max samples in a 16bit wave). But this is interesting. Going back to -20dbFS references- peak meter in SoundForge reads the same whether the wave is square or sine, but the peak sample values of both are the same.


I got similar values (-3278 and +3277) when I generated a -20dBFS test tone in Adobe Audition (sine and square). I’m not familiar with how SoundForge uses ‘reference’, so I can’t comment on what will happen when generating signal at full scale. In Audition, it will be the full range (-32768 through to +32767).

I hope this helps.
Good luck making sense of it.

If as you say -20dBFS is equal 0dBu and on your VU meter a sine tone reads -5dBVU/-3dBVU, then I would say that it sounds about right, but not quiet...
The reason is because 0VU is actually +4dBu not 0dBu. So you are missing a dB in your reading.
Did you round it up/down or something?

My mistake.

Wherever I mentioned a value as Dbu, it should have simply read dbVU, or just VU. Neither Vegas or SoundForge report values in dbu, just VU, and of course their peak meters (dbFS).

I appreciate your reply.

It's funny but with SoundForge configured to set -20dbFS to 0VU (and it mentions 0VU = +4dbu), I wind up with a -20dbfs sine reading -3VU, which I would assume would read 0VU for a sine wave, instead of 0VU for a square wave. It's RMS for sure, but appears to be offset if you consider a sine wave a proper reference signal.

Also that SoundForge peak meter doesn't give a dam about power, it reads the same dbfs no matter the wave shape.


I got similar values (-3278 and +3277) when I generated a -20dBFS test tone in Adobe Audition (sine and square). I’m not familiar with how SoundForge uses ‘reference’, so I can’t comment on what will happen when generating signal at full scale. In Audition, it will be the full range (-32768 through to +32767).

It's the same with SoundForge, it's just displaying sample values/1000. I just needed to see if SoundForge, when told to generate a wave, if both square and siine waves had identical peak values or not - and they did. I needed to know exactly what the peak meter was getting.

Now the kicker was that Vegas' peak meter isn't immune to wave shape - in contrast to SoundForge - it will register the power difference.

As far as Vegas' VU meter, it seems to respond correctly in RMS, it's just off by -5VU, and I cant figure out what sets its reference or if I can change it.