SuperG

bitflipper
3. Peak limiting is only indirectly related to perceived volume. It's entirely possible to achieve competitive loudness without exceeding a -1.0 dB maximum peak, or even less. Perceived loudness is all about average RMS, and when you lower your brickwall limit you raise RMS, so sometimes setting a lower limit will actually make your mix sound louder.

 
Good advice.  
 
I roughly adjust levels by eye using K-14 levels in Ozone 7. I'm not really interested in hitting that last 1 db. When compiling for a CD, I also check Ozone 7's 'integrated' meter setting to make sure the perceived loudness of all the tracks are compatible. For me this has been working out at about and average of -16LUFS.
 

Mister G,
 
If I were to type a post in this thread, it would be exactly what you typed. I also use Ozone7 and shoot for -16LUFS, (not participating in the "Loudness War"), use a Sound meter and finally my ears to make sure all the songs are roughly the same perceived loudness. I don't want to have to touch the volume control when I'm listening to one of my albums.
 
Walt
 
Walt

PeterMc
I would have thought it would be difficult to give precise guidance as to how to avoid intersample overs in the analog domain by reducing the digital peaks. It would depend on the end-listeners's individual DAC reconstruction filters, and also depend on the slope of the transients in your music. In other words, different DACs will interpolate between digital data using a variety of curve-fitting algorithms, and steep slopes in the music are likely to give more grief (it's harder to "turn the corner").
 
Having said that, while doing a little reading on this, I discovered that Ozone's maximizer has a little button labelled "true peak limiting". I'd always wondered what this did. Apparently it prevents clipping in the digital analog domain by being smart about the maximizing algorithm (and presumably making some assumptions about the end-users DAC). There must be other plugins out there with similar functionality.
 
Cheers, Peter.

You're absolutely right, whether or not > 0 dB results in analog clipping depends on whether the consumer's DAC has 3-6 dB headroom. Higher-quality players will. But the only assumption we can safely make is that a properly-functioning player (CD or MP3 or whatever) can handle 0 dB analog signals. We shouldn't assume that it can handle +3 to +6 dB, which many battery-operated devices can't.
 
You have answered rodreb's question, which is how to tell if your limiter can detect ISPs. Many limiters, including Ozone, offer a "true peak" or "ISP detection" option, or something along those lines. That simply means they're oversampling internally. Other limiters, such as Pro-L, don't call it that but instead give you a choice of 2x or 4x oversampling factors for essentially the same option. Concrete Limiter employs 16x oversampling.
 
We should also note that oversampling isn't the only technique for figuring out the "true" peak value, and some software such as Adobe Audition goes further, actually calculating it through algebra. But these are usually offline processes because they're so CPU-intensive. 

SSL's X-ISM is an inter-sample distortion meter. It emulates the roundoffs of smoothing filters and extrapolates whether anything goes above 0.
 
It's a discontinued piece of software (it was introduced 9 years ago) but is still available for download (and as far as I can tell, still works) on SSL's Japanese site. It does chew some CPU, though.

always -0.1 with a brick wall on the A bus

Anderton
SSL's X-ISM is an inter-sample distortion meter. It emulates the roundoffs of smoothing filters and extrapolates whether anything goes above 0.
 
It's a discontinued piece of software (it was introduced 9 years ago) but is still available for download (and as far as I can tell, still works) on SSL's Japanese site. It does chew some CPU, though.

Not found on the official site, but I found in http://store.solidstatelogic.com/sites/default/files/00bbada4-cc18-4c72-ad86-64e8bd511965/SSL_X-ISM_Setup.zip
 

@Craig
 
I never use that remove DC Offset tool. I'm thinking I should be ... ?
 
Should I be doing this on each audio track or only on bounces?
 
I just took a sec and analyzed a few tracks. Several had -100 levels or more... only one was less than -90. It was -85.
 
If I should apply it to all my tracks, can I select them all and apply it to all of them at once?
 
Thanks,

G
 

Hi Chuckebaby,
   I've read these posts and I think I have some useful advice.
1.  Dynamic range is the big gig in audio for loudness, recording, mixing, mastering and such.
Record at 24 bit (144dB S/N) and every consumer format will allow you to compress to a more useful, but inclusive, dynamic range.  This will allow you plenty of headroom and noise level removal.
2.  Try to edit all recordings to a -1dBfs level... you can always turn things down.  It is best to avoid turning up the levels as your noise will get louder too.
3.  Try to mix to a -3dBfs level... transients may compromise your mix levels but you can either manually adjust them, reexamine your source files or use a limiter when going for more loudness.
4.  When mastering, it is important to use the standards for presentation, i.e. DVD, CD, theater, surround sound, mp3, internet, ect...  There are no true standards to make "all-in-one" masters.
5.  Avoid generic mastering sites as they are just using hi-ratio multiband compressors to get the most volume and this will change your mix of dialogue (vocals), sound effects and music.
6.  If you master in 24 bit, you will have 144dB of S/N so you can convert to standard 16 bit presentations very easily.  I generally master to -1dBfs.  I try to keep the most important audio parts between -1dBfs and -12dBfs, depending upon the other material.  I generally keep secondary levels between -6dBfs and -18dBfs and third level sounds under -12dBfs and -24dBfs.  This generally will give you a 100% (loudest) to 12.5% (softest) volume.  This works well for theater film sound...  I compress a DVD sound level by 2:1 to go 100% to 25% (for normal... you may be required to reduce the dynamic range to 1/2 of that, depending upon the movie).  For music, it really depends upon the artist and style but most people like these above setting (hip-hop artist may like a 100%-50% setting to keep everything sound loud on ear-buds, iTunes, internet, ect.).  Any DNR lower than 25% may disappear in a "less-than-perfect" environment and any DNR larger than 50% will change the mix.
 
I hope you find this useful,
 
Tom
 
 
 

I used to do .3 until a couple of years ago, and after intersample overs became more public knowledge, I started setting my limiters to -1.0.  This makes it easier to save masters for both 24 and 16 bit wav and mp3 output files.  I don't really think I'll be losing much headroom peaking at -1.0 vs .3... and I haven't noticed any audible difference.

I am surprise at the amount of -0.3dB posts unless we aren't discussing -0.3dBfs.  I think people are misinterpreting dBfs, dBspl and dBu with dynamic range of audio recording, editing, mixing and mastering levels.  We need to keep in mind that a Decibel (1/10th of a Bell unit) is actually a ratio of two pieces of information... typically, this is a reference value of strength to a perceived value of strength.  In a sound file, this is the signal level compared, in a ratio, to the perceived volume... which also includes an ability to measure the dBspl of the presentation venue and the Fletcher-Munson curves to evaluate perception of the presentation venue.
 
If one masters to 0 thru -6dBfs, the resulting sound at theater volumes of 110dBspl will mean the entire sound will be 110dBspl to 104dBspl (this would be so loud you will get nauseous!)  Even lowering the DNR to 96dBspl will be amazingly loud in a film.  For a DVD, this would be fine... in an internet music presentation, this will mean some sounds are loud and some are too soft to really hear on a laptop speaker system.
 
I have found the best mix/mastering technique to allow for a 24dB DNR for modern sound.  A little bit of compression can the be used to master for various presentation requirements.  For some, specific requests, I have mastered to a 48dB DNR... but you can still compress at 2:1 to get back to 24dB DNR without compromising the basic sound.
 
 

What a fantastic thread. I'm definitely bookmarking it.
Great info.
Thanks to all for sharing your knowledge.
Ray (aka Oluon)