AAMS Automatic Mastering Software

Lately I’ve been through a transition. I’ve decided to open my mind and try out many of the modern tools and tactics that in the past I’ve avoided. Sample layering on drums has been a great example of a modern technique that has made my clients much happier. Recording guitars DI and then reamping them has been another. So even though I was apprehensive, I decided to give AAMS mastering software a try.

How AAMS Works
AAMS has the ability “scan” an audio file and save it as a “reference”. I don’t know the details, but it essentially recognizes the frequency response of whatever file you toss in it. I’m sure there is more going on than just frequency response analysis, but the general idea is to be able to clone the “vibe” of one track and then add that “vibe” to your track. This seems like a noble idea in theory. So does it really work?

AAMS In The Real World
I downloaded the free version (for a limited number of uses) and fired up a few of the various preset references and applied those to a mix I was working on for a local client. Each time the results were really bad. I said to myself “all is not lost”. I decided to scan a few different references from a few different albums that were in similar genres. After processing, I can’t say that I was any happier with the results. I didn’t feel like any of my mixes took on a similar “vibe” as the tracks I had selected as references.

Out of curiosity, I tried it on a dance music thingy I was making. This track consisted entirely of virtual instruments from MIDI land and I was very happy with the way the mixes sounded as is. After running them through various presets in AAMS, I was never happy. I thought that AAMS did a better job with the dance music stuff, but my initial vision simply was not respected by the software.

AAMS was way too aggressive. All the results reminded me of what I’ve seen numerous times in guys doing their very first mix on their 2bus. You can immediately hear extreme over equalization. There is only so much processing you can do after the tracks are mixed.

What Is Mastering?
I have to explain this so often and many people who’ve never even used professional mastering will argue to the death. Mastering is what you do to make a collection of great mixes sound like a cohesive album. Mastering takes care of more logistical type problems such as noise between tracks, amount of time between tracks, and overall volume level between tracks. Mastering is about minimizing the distractions from song to song.

Mastering is not about making up for inadequate mixes. It simply does not work that way. A mastering engineer has his hands tied when only dealing with the stereo mix. Mastering engineers don’t have the ability to adjust levels, manipulate eq, or manipulate effects on any of the individual tracks.

I’m not sure who created this myth that mastering will dramatically improve a mix, but they clearly have a different definition of “dramatic” than I do. It’s impossible for a mastering engineering to uphold the mixing engineer’s original intent and do something dramatic at the same time. Mastering requires a very soft touch.

AAMS Is No Mastering Engineer
I think AAMS has a major flaw. It’s ability to “analyze” a track is very poor compared to that of a skilled human being. The idea of scanning a square peg reference and cramming (Oh no! There’s that “cramming” word again!) that into a round hole is WAY too authoritarian for me. I used the phrase “square peg” because the processing that AAMS uses does not always fit what I believe is best for my mix.

Because AAMS is simply software, it has no way of knowing if I’m happy with my mix or not. When Eric Conn mastered an album for me, he asked me which mix best captivated the entire album so he could use that as his “reference” to decide what to go for on the entire album. (Remember, mastering is about tweaking individual tracks to have a similar vibe! It’s not about enhancing on a track by track basis.) Eric continually went back to this reference tune as he mastered more and more songs. It was brutally obvious that context was extremely important. Making each song “fit” in the puzzle was extremely important! He never once pulled out another artists mastered version to make my album sound more like that one. It just wouldn’t make sense to work in such a fashion.

AAMS goes through all lengths to make the analysis of my tracks match the analysis of whatever reference is selected. The only problem with “all lengths” is that it includes processing the track in a way that does not sound appealing. You see, sound quality is not really AAMS’s principle function. Improved sound quality is supposed to be a result of the analysis of processing, but that is very hit or miss.

Why Excessive Emulation Doesn’t Make Sense During Mastering
The entire concept behind AAMS is emulation. Emulation in and of itself is not a bad thing. It’s totally acceptable for a musician to have “influences” and in the end the emulating of all these influences is a big part of coming up with something entirely new. I can live with that. However, it’s one thing for a guitarist to have influences like Eric Clapton and Kirk Hammet. It’s another thing to boost 110Hz by 6dB and cut 90Hz by 5dB to fit some kind of curve from another recording.

It needs to be said that using 5dB during mastering is A ROBO TON of EQ!!! I have a feeling that no mastering engineer would ever get this aggressive. Eric Conn told me that if more than 2dB of EQ is needed it’s probably time for a remix.

In my upcoming home recording book, I make multiple references to the fact that EQ changes are not necessarily the same thing as tone changes. You can never EQ Boston to sound like Queen. You can not EQ Brian Setzer to sound like Napalm Death. You can’t EQ Cindy Lauper to sound like Norah Jones. You can’t EQ an American to speak with a British accent. It just doesn’t work that way. There is much more going on than EQ in the sound of a recording. I think we can all agree on this. So emulating the EQ of another album doesn’t mean we’ll sound like that album. In fact, the emulation can actually cause us to sound even less like that album.

For example, lets just say that in order to mimic the frequency response of X major label production, I have to boost 110Hz by 6dB and I had to cut 90Hz by 5dB. Does this mean my kick drum will have the same massive girth and my bass track will have the same thickness as the major label recording? Not at all!

The tuning of the kick drum, the way it was recorded, and the way it was mixed may be relying on 90Hz to provide the foundation for the kick. Its’ very possible that cutting 90Hz by 5dB will cut the heart right out of that kick drum. Maybe boosting 110Hz will do extremely little to bring back the life and impact that was sucked out. So just because the foundation of the kick drum is @ 110Hz on X big boy recording doesn’t mean that our kick drum will benefit. A real mastering engineering (or even a half-assed mixing engineer) would immediately pick up on this and would not make entirely excessive decisions that wrecks the recording.

I think some beginners could assume that their 90Hz kick drum is “wrong” and the 110Hz kick drum is “right”. (Note: these numbers are entirely arbitrary). That is completely not the case. If 110Hz was “right’ you would have tracked and mixed it that way. (If you are looking for mastering to do the mixes for you, think again!) We all have our preferences as to where we want to hear our kick drum meat. There are no wrongs or rights as long as the music is effective.

The problem here is our excessive emulation has led us to make decisions that made our music less effective. What worked on X production that sounds great did not work in our example here. It was meant to be that our kick drum was supposed to sit at 90Hz and now we have lost that.

The same goes for the bass guitar. The “center” of our bass guitar may be really deep or it may be much higher. You never really know. The low end on Fuel’s first album “Sunburn” is dramatically deeper and thicker than that of Guns N Roses “Appetite For Destruction” even though they both sit in the rock genre. There is no doubt about it. If you were to apply Fuels reference to any pre-mastered song on Appetite For Destruction, you’d end up with a dramatically boomier audio track. Appetite For Destruction never intended to have a big, thick bass guitar. It intended to have a nasty, distorted bass with lots of definition in the midrange.

Real Engineering vs AAMS
In my art masterpiece here it’s easy to see that with every engineering process there are two things that are required. We must have a source to deal with and we must have and we must have a vision as to what we want this source to sound like in the end. Then, using more of an Algebra-type of method we try to figure out what “X” is. We say “How do I take this source and get it to match my vision?”.

AAMS takes a different approach.

AAMS takes the source material, applies the processing based on a specific reference, and then spits out the result. There is no emphasis on “the vision”. Some may argue that we make the decision on what we want AAMS to sound like when we load a specific reference. Here lies the big problem. Just because we apply a James Taylor reference to our tune does not mean our tune will mimic James Taylor in any way. It’s supposed to, but in my experience it simply does not work as advertised because it’s too hard to predict wha that James Taylor reference will really do. It certainly doesn’t work without casualties.

The Search For Presets
I know that quite a few beginners in home recording have no problem with presets. This was discussed recently here on the Recording Review forum. However, what happens when we take the label off of all our presets and essentially have to blindly choose one preset or another based on listening? In the case of doing this with a compressor, it’s not that big of deal to hear if the compressor is doing what we want. In the case of AAMS, you have to process the file to really know. This can take 20 minutes or more.

In the time it takes me to find the most ideal AAMS preset I could have tracked, mixed, and mastered five albums. A person could search for months to find an AAMS reference that doesn’t tear up their mix. Why can’t a person just address mastering the old fashioned way even if the budget forces them to do it themselves. Import the wav file into your recording software of choice. Listen. If one mix has traits that stick out in a distracting manner from other mixes, address it the best you can.

You’ll find that if your vision in the mix has already been dealt with there really isn’t much you need to do, if anything. All you really need to do is make sure all the tunes have a similar, non-distracting tonality. That’s not overly difficult to do when compared to actually getting a song to sound exciting in the first place.

If You Can Find References…..
My only real gripe with AAMS is the fact that it’s hard to find references that actually make the mixes I’m working on actually sound good. I’ve already established that randomly grabbing a reference from a song in a similar genre is simply not good enough. There are too many differences that end up distorting my vision for the mix. It’s tough to find the time to try out multiple references. I tossed AAMS on my TV computer and it took 3 hours to process 6 songs in batch mode.

However, if you do manage to find references that actually work for your genre, AAMS does an decent job at evening out the differences. This is good if you totally lost your hearing in the war. If you lost your hearing in the war, you wouldn’t have made it this far though, son. If you are so incompetent that you can’t make a little .5dB cut at 120Hz, how in the hell did you track and mix and entire recording? Have more confidence in yourself!!!!

My Experiment
I had a theory that the best way to find a reference was to use the reference the big boys use when mastering. They choose the best sounding mix on the cd and use it as their reference. I did the same on a project I recently finished mixing. I picked the mix I liked best and made a reference out of it. I then applied that reference to all 6 songs. The end results was the best I’ve encountered yet.

I went ahead and mastered these songs the way I always do in Cubase. I made a 1dB cut @ 4k to one song. I put a multi-band compressor @ 2k on another song cutting a max of .5dB. I put a mult-band compressor up at 10k on another song cutting a max of .5dB. Done.

I think my manual version sounds better. AAMS did a dramatically better job since the reference I chose was suited for the mixes, but I think the fidelity of the manually mastered version is a hair better. If I was in an extreme hurry….well…AAMS takes so long to process that it would still be quicker to master it myself.

Conclusion
It’s hard as hell to find references that actually sound good with your mix. If you manage to find a good reference, AAMS can do the mastering for you in about 5x the time it would take you to do it yourself. If you can’t find references, you can waste months trying to find references that are compatible with your mixes. You can always create references with your own mixes, but if you are going to master one song, why not master the others.

Tags: AAMS, Automatic mastering

This entry was posted on Tuesday, September 2nd, 2008 at 12:23 pm and is filed under Mastering. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.