Mastering at -1dB ?

[spacef]

Hi all,

- Just to know, if any of you do mastering, do you use the -1dB peak value or do you choose something else ?
- Myself I aim at -0.40/0.30 dB approx, but it is arbitrary.
- I tried -1dB and found no real difference in level perception (the perceived loudness, the RMS etc, is the same). So I could go this way if useful.

What is the -1dB headroom supposed to do in practice ?

Thanks !

PS: I attach a pic of the numbers I am into (-9 / -8.5 Integrated RMS nowadays - I still have a l ot of difficulty going fully at -9dB and not trying to stay at -8 ).

[fra77x2]

no
you can hit 0dB for individual samples. A rule is to not have consecutive samples that reach 0 dB. One sample 0 dB is not considered clipping. Two in anrow or more it is.

[Rafe]

What application is the screenshot from?

Cheers

[spacef]

the screenshot is from Izotope RX audio editor,
as found in menu "/windows/waveform statistics".

About the -1dB True peak, I wanted to find out if it allowed better behaviour with DJ equipment (equalizer, bass/highs, filters etc). I have noticed a difference on an HiFi amp, but was wondering if it had something to do with that or wether -1dB is just a value "out of someone's hat".

[valis]

0dB can give 'inter-sample peaks' on consumer converters. -0.03dBfs is supposed to prevent this, but years ago TC Electronic did a great whitepaper showing most consumer conversion circuits don't have the headroom to operate at near peaks, a clean waveform passed to the analog circuit (no ISP occurring) would still clip the output stage. If -1 sounds better to you, that's likely why.

So something under -0.3 dBFs was shown to have much less distortion. Even with the loudness war in the rear-view mirror, we still are capable of dynamic control well beyond what we did back then, since we can slap as many processors into our software (and DSP) chains as we want now.

All DJ software now also does relative gain (like Foobar2000's "replaygain"), as do the better software audio playback implementations (thanks to the hydrogenaudio community and Foobar2000). Whether this means that your -1 will be passed through without being re-clipped by the user's settings is unfortunately up to whether or not they respect the default settings when doing gain detection and waveform building (pre-scanning tracks). Serato has the best overall gain-setting approach, Pioneer's is very specific to which mixer you're using, and Traktor/VirtualDJ etc are the most easily abused.

[valis]

Just for parity's sake, I'll back up my statements.

Hydrogenaudio forum discussions, Head-fi contributors and AES whitepapers are representative of the purchasing audience and producers of the tools alike, and so we react to and document the tools we purchase.

Level Control in Digital Mastering By Søren H. Nielsen & Thomas Lund
(free link: https://service-tcgroup.tcelectronic.co ... 109(1).pdf)

https://secure.aes.org/forum/pubs/journal/?elib=3393
(Google scholar link: https://scholar.google.com/citations?vi ... pCki6q_DkC)
( Full list of his contributions from Google Scholar: https://scholar.google.com/citations?us ... AAAJ&hl=en )

https://www.head-fi.org/threads/dac-and ... st-8383293


Ringing Enforcement. Keith Howard Hi Fi News Jan 2002 p72

https://www.audiomisc.co.uk/HFN/OverTheTop/OTT.html (this only applies to waveform reconstruction and doesn't touch on the analog circuit's ability to track that waveform and the possibility of calibration of that circuit leaving no headroom, but it has pretty pictures).

I will underscore again that some DACs (like the CS4398, which clips near +2 dBFS) have headroom for them to handle even ISP's, but the entire circuit (from a digital filter's contribution to ripple, to analog reconstruction and headroom) need to be considered when it comes to headroom.