Overcooked ?

[Thalamus]

Hi there - just been harbalising some of my old tunes ..

I have found that when I make adjustments in HarBal that sound fine the output file is clipped and produces nasty clicks in some cases - I was wondering why it should sound fine within HarBal and not good in the output file .. I have not touched the gain control?

Is this to do with the 'peak' indicator?

[har-bal]

Hi there - just been harbalising some of my old tunes ..

I have found that when I make adjustments in HarBal that sound fine the output file is clipped and produces nasty clicks in some cases - I was wondering why it should sound fine within HarBal and not good in the output file .. I have not touched the gain control?

Is this to do with the 'peak' indicator?

Thalamus

I will paste the answer to your question from another one of our users.

Due to the nature of filters used in Har-bal, clipping can occur if you try and tweak a file that is already close to saturation (0dBfs).

If you bare this in mind and make sure your source file is not already normalized and/or heavily maximized to 0dBfs, then you should have no problems.

In short, record to a healthy level, mixdown to a healthy level, Har-balize if required as part of the mastering process and then finally maximize using a high quality maximizer/multi-band compressor limiter.


Earle

[Thalamus]

I see...

I was just interested in the technical difference between listening to the file in HarBal and the exported file - non of the clipping is aparent in the audio during preview where (I assume) the filters are working - just in the exported file.

[har-bal]

I see...

I was just interested in the technical difference between listening to the file in HarBal and the exported file - non of the clipping is aparent in the audio during preview where (I assume) the filters are working - just in the exported file.

Yes, that is correct. Anytime I load a file into Har-Bal and the saturation meter starts registering I automatically pull down my output fader until it disappears. I then restore loudness later with a limiter.

Earle

[Thalamus]

Earl, I can see the workarounds but am looking for a reason why the clipping does not happen during audition but manifests in the output file.

I'm just curious as to the difference between the two?

[har-bal]

Earl, I can see the workarounds but am looking for a reason why the clipping does not happen during audition but manifests in the output file.

I'm just curious as to the difference between the two?

Thalamus

Let me make sure I understand you completely.

1. You load a file into Har-Bal
2. Without making any adjustments does your saturation meter register? Y/N
3. If not, are you hearing any clipping Y/N
3. The clipping registering after you make EQ adjustments only? Y/N

I am assuming that after you answer these questions you are loading a file into Har-Bal and NOT hearing any clipping and NOT seeing any activity in your saturation meter.

Yet when you record the file your are hearing cliiping

If this is the case I will have Paavo contact you. The only case of clipping we have had is where the inputted file was actually too loud and the volume needed to be brought down.

The only other way this could happen is if you are reading a wave file directly from a CD on your machine and are experiencing jittering.

Earle

[Thalamus]

I think in the cases where bad clipping has happened there has been some activity on the saturation meter.
How is this meter calibrated? Does just one green square = an unacceptable overload? or does it have to go into the 'red' before clipping is happening.

My main question is quite simply this :

Even when the saturation meter is showing some activity I hear no (zero) clipping in the audio ... it is only when I export the file that the clipping becomes aparent as audible problems. Why is that? Why does it sound fine (regardless of saturation meter activity) in HarBal and not fine on export?

[dbmasters]

Let me toss this out there to, there are times, when I start tweaking during playback, my files start clipping and distorting and won't stop unless I click stop, then start again, then it stops.

[HarBal]

I think in the cases where bad clipping has happened there has been some activity on the saturation meter.
How is this meter calibrated? Does just one green square = an unacceptable overload? or does it have to go into the 'red' before clipping is happening.

My main question is quite simply this :

Even when the saturation meter is showing some activity I hear no (zero) clipping in the audio ... it is only when I export the file that the clipping becomes aparent as audible problems. Why is that? Why does it sound fine (regardless of saturation meter activity) in HarBal and not fine on export?

Thalamus,

Theoretically there is not reason why this should happen as the filtering done in both cases is identical. However, I have noted that the LIBSNDFILE library inverts samples that are full scale in the positive direction, converting them into negative full scale values. I noted this when doing some impulse response tests on my latest filter engine changes to verify that is was functioning corectly. If this is happening in your case it would certainly make clipping clearly audible. I'll have a look at figuring out whether the inversion occurs on reading or writing and then report the issue to the author of LIBSNDFILE to see if he can fix the issue. Thanks, for pointing this out.

Regards,


Paavo.

[electro]

Doesn't HarBal produce a 32bit float file, if it's opens a 32bit float file?

[HarBal]

Electro,

That isn't the issue. The problem is a saturated positive value gets erroneously converted to a negative one (or the other way around). It's a bug in the software and from memory, it happens in LIBSNDFILE.

I was supposed to have verified this and reported a bug to the author but I haven't done so yet. Thanks for reminding me. My guess is this bug only shows up for floating point format files but I haven't verified it.

Paavo.