amplitude statistics and unusual ways of mastering…perceived loudness v peak amplitude v the human ear

because I sometimes slightly mistrust my “50-something”-year-old-ears, and, because, by coincidence, I happen to have 26 days left in my latest trial of adobe audition :-), I’ve decided to take advantage of a feature of adobe audition that is very useful – an amplitude measurement tool that you can run on any audio track, that describes it’s sonic characteristics in terms of “loudness” – obviously, it gives you left and right channel decibel readings, in peak amplitude and a lot of other flavours, too;  …sure, that’s the basic thing you want, but it also provides a large number of other supporting figures, including one that appeals to me personally – “perceived loudness”.  not to mention a single overall perceived loudness figure that seems to be based on to a british standard…

I’ve always had a soft spot for analysers (or “analyzers” as I grew up calling them in california) – I worked in an electronic test equipment repair and calibration laboratory when I was, uh, younger…so am familiar with a host of meters and analysers, but these kinds of software-based tools (a very capable measuring device in a very, very capable audio application!) have come a long, long way from the test equipment that I worked with for 12 years some 30 odd years ago, and adobe audition has a seriously detailed tool for assessing the levels, expressed in several different ways, of any audio track.

I’ve been listening on headphones to the completed tracks for “gone native”, and, by my “ear” – the “perceived loudness” seems reasonably even between all 19 tracks.  interestingly, however, in order to achieve this, I’ve done something quite unorthodox.

however…I’ve always been unorthodox when mastering CDs, and my peers would look in genteel horror upon me when I would commit the almost unforgivable “sin” of mastering an entire album to – 6 db (rather than the more traditional “just under zero db” that most engineers shoot for) – for example.

so – master up to zero db like most people?? not dave stafford – he wanted it quiet, not loud.  and, maybe…for a totally ambient album made in the mid-1990s – that’s a reasonable concept, it actually works OK, instead of having to turn the disc DOWN, listeners had to turn it UP – revealing more detail – and, preserving the dynamics better for the listener (in my opinion).  if it’s already as loud as it can go – they can’t turn it up – whereas with some headroom – at minus 3 db or lower – they can turn it up quite a bit… but it’s all subjective, and every pair of ears perceives sound differently anyway…

I am old-fashioned, and when the trend of seriously compressing the hell out of your CD tracks, and having them not just hitting 0 db but possibly peaking above that, the practice of making rock CDs in particular really “in your face” began – well, it’s just not “me”.  since most of my CDs were and are ambient, mastering to less than 0, whether it be as incredibly low as minus 6 db, or, in more recent years, closer to minus 3 db – it just worked for me – I just don’t get near zero decibels – it’s not “the sound” for me.

but when it came time to make “gone native” – well, it is mostly a rock or prog rock album, therefore, it should be louder than my ambient work, so I decided I would target minus 3 db as my baseline “loudness” for this record for the rock tracks – that seemed obvious.  however, in working with some of the tracks on “side two” (tracks 13 – 19), I found that to “match” one track to another that was measuring as minus 3…that I had to make the track I was matching quieter still, in one case, minus 10 db !!

things got even more interesting when I did the measurements using the tool.  I had three tracks that had a very small imbalance between right and left – so I adjusted those.  (although, I do happily allow a small amount, say, half a db at the most, of “imbalance” if that’s what it takes to make the track sound right – I do not insist on left and right being totally equal).

I actually created a couple of alternate “versions” of the master, so I could trial having the last 7 tracks significantly quieter than the the first 12, or…closer in both peak and perceived loudness to the first 12 tracks.  various listening sessions eventually proved out which concept works best overall for this somewhat unusual album…

having both the perceived loudness figures and the single, overall figure is helpful with this, too, but no matter how you measure, or listen,  my “measured perceived loudness” compared to “ear perceived loudness” tests have proved very interesting…

in the end, I decided that the peak amplitude figures – while important, well, I could not let it worry me that the way it the sequence of songs turned out, that some of the ambient tracks did indeed end up with peak amplitudes closer to minus 10 than minus 3.  that is not traditional, normally, you would master all the tracks close to zero db, or, master all the tracks to minus 3.  I’ve done something completely…other 🙂

normally, I would have thought this wide range of levels would be a problem – but in the end, it’s actually not – because:  most of the 12 loud songs, which vary perhaps between minus 2.5 and minus 4.5 db, all have roughly a perceived loudness of…minus 18 db !

so when it comes to the quieter songs, where I gradually lower the level of the tracks as the album moves towards it’s close…even though a couple of the tracks end up at minus 10 db peak amplitude…their effective perceived loudness drops much more slowly – so in the land of peak amplitude, the loud songs may start at minus 2.5 db, and slope down to minus 10 – so a span of about 8.5 db – however…in the land of perceived loudness, the loud songs start at minus 18, and slope down to minus 25 – a span closer to 7 db.  either way, mainly because of careful selection in the ordering of the last seven tracks, the ear is “tricked” – and it doesn’t notice the somewhat gradual (although not necessarily linear!) descent from roughly minus 18 db down to as low as minus 25 db – in perceived loudness (minus 10 db in peak amplitude) – all figures approximate 🙂

or, in perhaps clearer language, the album gradually gets quieter as it moves from the 12 loud songs, into the 7 ambient / experimental songs, but, the perception is of one continuum – and you just don’t really hear that it gets quieter – sort of – towards the end of the record.

I did an experiment, one of the variations I created, where I boosted the last seven tracks to “match” the first 12, so they all had roughly the same perceived loudness – about minus 18.  It sounded…terrible!  the tracks seemed harsh, too loud, and it just sounded completely wrong to my ears!

so – right or wrong – or rather, by all the rules – wrong – to my ear, the running order I currently have, just “works”.  my test audiences didn’t report any issues with this “gradual drop” method either, so hopefully most listeners will never realise (unless they read this of course!).  I would say – I’ve mastered a LOT of tracks now, over a lot of years, and you can take two songs, and run them through identical processes, master them identically, target zero db – and still, when you play them back – one sounds MUCH louder than the other – it’s inexplicable, because even with these sophisticated tools – still, still – the human ear is the best judge, because the numbers do NOT guarantee consistency in volume – only the human ear can do that.  or so it seems to me…

and if you just listen – you would never, ever realise all this measuring and comparing was done – in fact, really, in one sense – it didn’t need to be done, because in the end – I still trust my ears over what the numbers tell me – the measurements just help me understand better the variation in perceived loudness, rather than always relying on the standard peak amplitude that most people “go by” – and I am sure I will continue to use this to analyse album sequences for a long time – a fantastic tool – and props to adobe for creating adobe audition, which I also love for it’s click and pop removal qualities (guess what I will be doing for the next 25 days….)

happy mastering – and, don’t try this (the above) at home 🙂

but the best news is, no matter what the measured values say, in listening to the test pressings – it just sounds really “even”, and really good – if I do say so myself – so I am very, very pleased indeed.



Please leave a reply.

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s