RMS and Peaks in Har-Bal

Having problems using the greatest Visual Mastering software of the century? Use this area of the Forum to post your technical questions to Earle and Paavo regarding Har-Bal or ask questions regarding how to work on a certain area of the software? Post away!
Post Reply
petr
Posts: 17
Joined: Thu Jan 17, 2008 8:31 am

RMS and Peaks in Har-Bal

Post by petr »

Hello Earle and Paavo,
I do not understand one thing in HB. When I load for example an audio file to HB there is a difference between the information in Sound Forge (RMS -13, peak 0) and the HB where I can see for example RMS -8, peak +5?!) What does it mean? Is it OK and I shall check the numbers that -8-5=-13 RMS? So why there is the peak in HB written +5 and not 0?) Does it have a sense writing it like you do? Does it tell me the information to help me this way?

I also saw a strange thing that for example when I matched the reference file with my file to be remade and there were informations like this.
Sound Forge: reference file: RMS -8, File to be remade: RMS: -12
HB: When I put MATCH RMS: the HB limiter says for example only -1 and not the -12-8=-4?!

Thanks for your reply.
Cheers.
Petr
petr
Posts: 17
Joined: Thu Jan 17, 2008 8:31 am

NOW EXACTLY

Post by petr »

NOW EXACTLY - I AM READING AS FOLLOWS:
MY SIONG XY.WAV
SOUND FORGE -13,2 RMS 0 PEAK
HAR-BAL -10,55 AVERAGE +4,30 PEAK

THE A FILE TO BECOME THE REFERENCE:
SOUND FORGE -11,4 RMS -0,10 PEAK
HAR-BAL -8,48 AVERAGE +6,02 PEAK

I MAKE THE REFERENCE FROM THE REFERENCE WAV FILE AND DO THE MATCH LAUDNESS AND THE LIMITER GAIN SHOWS ME +0,40 CHANGE

THERE SHOULD BE -13,2 (MY FILE XY.WAV)+ 11,40 (THE LEVEL OF THE REFERENCE) = -2 OR I DO NOT UNDERSTAND IT?!
HarBal
Site Admin
Posts: 761
Joined: Mon Apr 19, 2004 8:18 pm
Contact:

Post by HarBal »

Hi Petr,

The average levels reported by Har-Bal look consistent with Sound Forge in that they are different by 3dB. This difference arises from the fact that the reference level for Har-Bal is a full scale sine wave whereas the reference level for Sound Forge is a full scale square wave. They differ in power by 3dB.

Although the full scale square wave reference is in common use it is not the recommended standard, which is a full scale sine.

The peak value reported by Har-Bal is not something that is directly comparable with sound forge. It is more an illustrative figure than a particularly useful one and perhaps I shall remove it in future versions. The reason I say this is because the peak spectrum is merely the aggregate of peak values across the spectrum (ie. the maximum values) which do occur at different times in the track rather than simultaneously. That is why it is possible for the peak figure to be greater than 0dB, which is seemingly impossible until you realise that the spectrum components don't all occur at the same time.

On loudness matching, the perceived loudness is more complicated than just the average level, so you'll find tracks where the optimum levels as far as perceived loudness, will not be the same as loudness matching will guess. If the genre and intensity of the tracks is similar then loudness matching will work fine but if you are trying to match loudness of a ballad with a rock song then you'll need some manual intervention as I'm not aware of any automatic process that can handle that situation effectively.

Cheers,


Paavo.
petr
Posts: 17
Joined: Thu Jan 17, 2008 8:31 am

So what shall I use to match the loudness?

Post by petr »

Paavo,
thanks for your explanation. So what shall I use in order to match the loudness?
I have a reference track I like and want to match.
I do the loudnessmatch and the limiter tells me a number to correct.
Shall I put this number into my final limiter (in the case I do not do anything else with the song after har-bal) or shall I compare the loudnes in the Sound Forge and correct according to the RMS from Sound Forge? You yourselves admit that there is a difference between these two programmes approx. 3dB. Which way would you recommend?
PS:
If I understand properly you explanation about the 3 dB means that when Sound Forge tells me that RMS is -13, the RMS measured by HB would be approx. -10 dB?
If you recommend us to keep the final mastered RMS between -11-13 dB, we sould use the values from HB and not the ones from SF?
PS1:
And if the HB tells me that my song does have the PEAK a + value (for example my +4,30), does that mean CLIPPING (!!)for me?! Shall I remix my song in order to achieve max. 0 dB PEAK in HB before using the HB?
HarBal
Site Admin
Posts: 761
Joined: Mon Apr 19, 2004 8:18 pm
Contact:

Re: So what shall I use to match the loudness?

Post by HarBal »

petr wrote:So what shall I use in order to match the loudness?

You can use the loudness matching tools in Har-Bal and if they sound right by all means go by them, and if they don't then do it manually. That is easy enough to do in Har-Bal simply by loading your reference track as a Har-Bal reference and then pressing toggle reference while playing back your track. If the loudness is comparable then you should hear a consistent volume between playing your track and the reference.

If you want to use an external limiter and/or your reference track has an RMS level more than about -10dB I'd suggest applying a gain cut filter to the reference to bring it down to -13dB or so and then match loudness using the above approach. Then the gain figure to apply to your limiter will be the difference in gain slider setting between the reference and the track filter.

petr wrote:I have a reference track I like and want to match.
I do the loudnessmatch and the limiter tells me a number to correct.
Shall I put this number into my final limiter (in the case I do not do anything else with the song after har-bal) or shall I compare the loudnes in the Sound Forge and correct according to the RMS from Sound Forge?

Don't use RMS levels at all. Use your ears and toggle reference as pointed out above.

petr wrote:You yourselves admit that there is a difference between these two programmes approx. 3dB. Which way would you recommend?

Neither. Loudness match through listening! The tools in Har-Bal are an aid, not a final solution. Use your ears.

The average RMS level of a track is just a figure of merit. Whether it be relative full scale sine or full scale square is not significant provided you know which is which. dB is a relative scale so you need to know what it is relative to, otherwise it is meaningless in conveying an absolute level.

The average RMS level of a track isn't a worthy figure of merit for much as far as I'm concerned. I'm really concerned about how it sounds. For me it only serves as a good indication of the bad practice of partaking in loudness wars. If the level is over about -11dB (relative sine) most recordings start sounding degraded to me. That's about all it indicates to me : when someone has slammed a recording!

petr wrote:If I understand properly you explanation about the 3 dB means that when Sound Forge tells me that RMS is -13, the RMS measured by HB would be approx. -10 dB?
If you recommend us to keep the final mastered RMS between -11-13 dB, we sould use the values from HB and not the ones from SF?


I'm talking figures relative to full scale sine (ie. Har-Bal) so in sound forge terms I'd be reticent to go for levels any higher than -13dB (-10dB in Har-Bal).

petr wrote:And if the HB tells me that my song does have the PEAK a + value (for example my +4,30), does that mean CLIPPING (!!)for me?! Shall I remix my song in order to achieve max. 0 dB PEAK in HB before using the HB?

The peak figure of merit tells you no such thing as I pointed out in my previous reply. That figure won't tell you if and when clipping will occur.

Cheers,


Paavo
petr
Posts: 17
Joined: Thu Jan 17, 2008 8:31 am

Thank you

Post by petr »

Thank you Paavo!
Cheers!
Petr
Post Reply