If you are here from yesterday’s post about recording this song, I will warn you that today’s post is going to get a bit technical and geeky. That’s because today I’m going to talk about the mixing process and then creating the music video you see above.
First, however, a little background on the tools, because I’ve made some changes recently. Many years ago when I was first deciding what software to use for producing music (and they are legion) I followed some sage advice and simply used the same software my best friend and collaborator had selected. At the time Sonar was at version 7. I bought the stripped-down Sonar Home Studio and selectively upgraded as specials were offered until I was up to the flagship Sonar 7 Producer. It worked well – so well that I felt no need to upgrade through five versions! Finally this Fall I took the plunge and upgraded to the Producer version of Sonar X3 (essentially version 12). Cakewalk had made numerous updates to the software in the intervening versions, and one of the things that I did was determine to use only the stock plug-ins that came with Sonar on this mix. Please note that I did use third-party instruments in the recording – I didn’t limit myself to just the instruments that came with Sonar.
The first thing immediately obvious to users of previous versions of Sonar is the brand new ProChannel. This is the strip immediately to the right of the main channel strip in the screenshot on the left. Sonar X3 comes with a half-dozen modules for the ProChannel (and you can buy a dozen more), and you will see several of them of the course of this article.
The top module is the new quad-curve EQ. This is a four-band parametric EQ with additional high- and low-pass filters. It’s cool because is has an RTA display behind the EQ curve visualization. This lets you see the frequencies of the incoming signal and I’d been wanting one of these for a while. It also has four “types” which change the way the bands overlap and how they respond the the “q” width control. On the lead vocals I filtered out the low frequencies and gave a little bump at 400 hz to counteract the 400 hz cut I made in the 2-buss EQ (see below). This 400 hz is where the “best” part of the voice lives – the part that gives it the most intelligibility. By cutting everything else at 400 hz it leaves room for the lead vocal to be more present and understandable.
The second module is the compressor, which I ultimately turned off after trying it for a while. It is an optical-type compressor, which I often like for vocals. Rather than a threshold control, you turn up the input gain, driving the signal into a preset threshold (which reacts somewhat dynamically to the material coming in). However, I wanted to control the threshold manually so I could better control the overall dynamic of the signal. I ended up inserting the Nomad Factory Fairchild clone. The controls are similar with the addition of a threshold control which allows changes in dynamic control without the automatic increase in level.
The last module is the console emulator, set to the Neve. For the uninitiated, this class of plug-in exist to add the subtle harmonic coloration that a high-quality mixing desk imparts to the signal. Some very serious analog proponents have called these plug-ins, specifically the Slate Digital Virtual Console Collection, a game changer for in-the-box mixing. However, a review I read of VCC versus the ProChannel console emulator that found the VCC clearly superior. To be honest, the effects are too subtle for me to make a judgement at this time, but since I purchased VCC prior to owning Sonar X3, I’ll probably go back to using it.
The guitar in this recording is an older version of RealGuitar by MusicLab. I really like RealGuitar since it can switch fairly seamlessly from finger-picking to strumming, and the picking can easily play just the chord tones and include authentic hammer-ons and pull-offs. Since the instrument is playing a supporting role in this production, I rolled off quite a bit of the lows and some highs. There is a slight presence bump around 1260 Hz which helps the remainder of the signal cut through the mix.
The compressor is set to knock off as much as 6 dB with a moderate attack and release. The moderate attack allows the initial transients of the strum to come through uncompressed, but fairly quickly get the sustain portion of the sound out of the way of the rest of the mix. At the bottom of the channel is the console emulator, this time set to the Trident A-Range emulation.
Solo violins tend to require the most processing no matter how well they’re recorded, especially when they’re expected to sound sweet and warm. They do cutting and bright really well. In big groups they shine at lush and warm. The EQ is to tool of choice for this transformation. I started by filtering out the low end. This is pretty much the only EQ move I make in solo. The trick is to raise the frequency until the sound begins to thin out and then drop it back just enough until it fills out again. This gets rid of rumble that can screw up your buss compression and do weird things with the energy in the track but keeps from making the whole mix thin and weak.
The rest of the EQ is a high shelf to get rid of any residual screechiness and a little bump at 335 hz to warm up the tone. The compressor is taking just a few dB off to even things up and it is followed up by the console emulator set to the Neve type.
Mixing the drums was expected to be the most involved but it really wasn’t too bad. I didn’t go nuts on the EQ for the individual drum tracks. The overheads are processed through a Mid/Side decoder to turn them into a stereo pair, and the buss also has a high-pass filter to remove the lows that aren’t necessary with a little dip at 400 hz to make room for the lead vocal.
The kick drum is rolled off at the top to filter out weird high frequencies that the kick mic picked up but don’t contribute to the overall sound. The snare just has the 400 hz dip.
The green track is what’s left of the original scratch drum track. When I programmed the drums I didn’t know whether we would be getting a live drummer or not, so I did the best I could. The sample library is Addictive Drums 2 which comes bundled with Sonar X3. It’s a fantastic drum library, but I find the MIDI patterns a little lacking if your aren’t doing straight-ahead rock. In the end I had to play a lot of the parts in by hand on a keyboard. Some of what the drummer played didn’t quite fit the concept for that section of the song, so I supplemented with the original track – best of both worlds. I think I ended up using about a minute and a half of the original drum track – some in place of the live drummer and some supplementing.
Everything is then run through the Nomad Factory stereo buss compressor taking off just a few dB to glue everything together. The compressor can be switched between peak and RMS mode – peak will control the attacks and RMS will even out the overall level. I’m in RMS mode here.
In sticking with my commitment to use only stock plug-ins from Sonar, I explored the included Breverb2 algorithmic reverb. I haven’t decided if I really like it, yet, although at first blush Valhalla Room is still my favorite. I searched through the presets, though, and did find one that stuck out as just what I was looking for for this production. Called Warm Orchestra it is a really sweet, fairly long reverb that melded with this ballad just perfectly.
I switched the ProChannel around so that the EQ was placed after the reverb, and filtered the lows and shelved out some of the highs. It is the rare reverb that doesn’t take up too much space in a mix without these adjustments.
The mix buss traditionally receives the most processing in my mixes. This method has recently been promoted by recording teachers Graham Cochrane and Joe Gilder, but this is the way I’ve been doing it for a while. The EQ has the final 400 hz cut (the one counteracted by the 400 hz boost back on the lead vocals) and the compressor is just shaving a dB or 2 off the top for the glueing effect. The console emulator is set to Neve and I’ve added the tape emulator to the end of the ProChannel chain.
The multi-band compressor is based on a tip I just picked up from Graham at the Recording Revolution, and he just picked it up at the NAMM show from mixer Greg Wells (Katy Perry, Adele). The idea is to just compress the mid bands (cutting off at 140 hz, 2 Khz and 6.5 Khz). This compression and resulting increase in level by just 2 dB adds a perceptible energy to the mix without noticeable increase to the overall level. I tried it on this mix and liked it so much I will probably employ the same trick on all my non-classical/jazz mixes in the future. Heck, I might try it on those, as well, just to see if I like it.
At the end of the chain is my Boost11 limiter. You can see quite a bit of level gain is being used to bring the final mix up to production levels. I ended up automating the Boost to bring the beginning and ending of the song up a little bit – the middle was just a little too much more loud than the ends. Just for fun I put together a composite screen shot of the entire project, just like they do in those Sound on Sound articles:
I’ve also shown the automation lanes for each track that uses automation. The lead vocal at the top has two sets of volume automation – the first is ducking the breaths at the beginning of each phrase. The second is setting an overall level for her vocals in the mix. The backing vocals in the second track have an automation lane for volume, and the second lane is the reverb send. I changed the reverb level for the section in the middle where she has some solo ad libs that needed to cut through better.
The piano, pad and guitar all have smaller levels of volume automation to get them to sit appropriately at various points throughout the mix. The violin has much larger amounts of volume automation because at various times it is a featured counterpoint to the melody and later more of a background instrument providing additional motion.
Finally, the buss automation at the bottom shows the change in overall choir level in the second track and the automation of the limiter boost I mentioned earlier on the very last track (green line).
Almost done. This is the first time I’ve ever produced a tracking video. All my previous videos have either been slideshows or scrolling sheet music videos. I knew that I wanted to do some picture-in-picture (PIP) with this video, which ruled out my old go-to, Windows Movie Maker. I did some research and landed on CyberLink PowerDirector and I’ve not been disappointed. At the top of the screenshot are the various clips that I imported into the project, and right of that is the preview window. Underneath is where the action is – the timeline. I only needed so many tracks because in the instrumental section where the dancers appear I wanted each new dance vignette to appear on top of everything else. Since the lowest track on the timeline always appears on top I needed a new track every time for each new scene in that section.
Even with all the PIP, motion and video correction I did with this video, I barely scratched the surface of what PowerDirector is capable of. It’s possibly the best $70 software purchase I’ve ever made. The biggest lesson learned, though, is to record EVERYTHING! Even PowerDirector can’t show footage that was never filmed!
So, that’s everything. If you’ve made it this far and you’re not a recording wonk like me (I’m looking at you, Will), congratulations! Now go back and enjoy the song and video again.