Har-Bal Users Forum BBS

Har-bal is giving me errors with 32 bit float files.
A wave shows a -13.08 average with +5.78 peaks.
The same file analyzed in Wavelab shows -13.06 ave but -0.10 peaks, which is correct.
Adjusting the eq in Har'bal results in clipping.
I'm using v 2.3r3.

/dave

Can you isolate a short section (file size of 1MB or less) of the track that contains the problem areas for Har-Bal and then email it to support@har-bal.com .

Thanks,


Paavo.

Thanks, Paavo.
I'm re-mastering the whole project, so I'll try the new files in HB and report back.

/dave

Hi Dave,

It might pay to be aware that the HB meterage is of power , not sample reading ... and the figures you cite seem not unusual ...
Your clipping ( if you're reading it from HB's meter ) could happen in any event where a boost is filtered in . HB has a clamped output of 0.1 dB ....... it wouldn't take much to kick somewhere in your 0.1dB file , over this .... Simply pull the gain down ... to remedy ...

Cheers ,

Evan .

dmac,

I see what you are talking about now thanks to hitmakers comments. The +5.78 peak power reading on har-bal is not an indication of clipping. that value can and does go above zero without any clipping occuring. As an explanation as to why, consider this.

You have a track in which you play 10 loud notes (ie. close to clipping) one after the other and each with a different pitch. If you were to analyse that file with Har-Bal the peak spectrum you get is the aggregate of the peak power of all those notes rolled into one. Now if you happen to play all those 10 notes simultaneoustly with exactly the same intensity you would get basically the same peak power spectrum analysis (assuming there were no clipping). However as each not individually was close to 0dB it doesn't take much imagination to realise that played together they would exceed 0dB. This is what you see in the peak power spectrum figure of merit and from the above you should be albe to see that if is possible to have a figure greater than 0dB without clipping.

Regards,


Paavo.

Thanks, I appreciate the replies.

The clipping I referred to was in the *._eq files created by HB, not while playing the wave file I was working on.

I've since noticed a major mistake in my processing chain before I got to the HB stage - I should have noticed this while Har-Balling, so I've put myself on probation.

I'm back re-mastering the songs, then on to HB to fine tune the EQ.

There may be a problem with my mastering chain - I apply HB after my master compressor (mainly because I can't insert HB in the fx chain).
I always set the limiter slider in HB to -0.6 on the assumption that will keep the output file under 0 db.
But on the work I referred to, the Har-Bal files were clipping.

I'll report back with the results of HB on the new files.

/dave

Hi Dave,

You could use the 'limiting' meter , down the bottom of HB , as a guide . If you set your gain such that no green bar is seen in it ( which is difficult because it gives no indication of approaching overs .... just when it is ) , your file will be ~0.1dB ...

Cheers ,

Evan .

Very useful tip.
Thanks,

/dave