We make a 'live' band recording in the studio, and mix and master the results all in one day!
This month's Session Notes is a little out of the ordinary, as my brief was not to make the perfect recording! Instead, I was invited to capture the 'warts and all' feel of the Malvern Hillbillies — a band who predominantly play live — and both to record and to mix a lot of their songs in a very short space of time. More specifically, they told me that they should sound like "a live band playing in a pub after a few beers”! So this was, in fact, less like a typical studio session and more like a live gig recording — though obviously one without an audience. By the end of the session, we'd recorded 13 tracks in around two and a half hours, and we mixed them all in the course of the afternoon, and as many readers will from time to time find themselves tasked with achieving a similar thing for a band, I thought it would be helpful to tell you how I went about it.
Utilitarian Recording
The location for this recording session was the multipurpose utility room that adjoins my home studio. I've installed a wall box with mic and line tie-lines in the room, and it regularly doubles up as a live room. There's a tiled floor and the room is surrounded by shelves of domestic products, spare guitars in cases, pet-food containers and myriad other odds and ends. It also contains a washing machine, a tumble drier and a small freezer, the last of which is switched off when recording. (Thankfully I've not yet forgotten to switch it back on again after recording!) Despite the basic and very 'domestic' setup, the random nature of the items stored in the room actually seems to result in a very nice acoustic for recording stringed instruments (I like to think that all the cans of food and bottles of bleach serve as effective diffusors), and those who've recorded in this space include Bert Jansch, Nigel Kennedy, Gordon Giltrap and Rick Sanders — so there must be something right about it!
Everyone played together, so little was really needed by way of artist monitoring, though Tony, the guitarist, asked for headphones so that he could hear his instrument and playing a little more clearly. All the recording was done using Apple Logic Pro X, with a Presonus AudioBox taking care of audio interface and mic preamp duties. After locking the cat flap to discourage unwanted visitors during the session, we made a start...
Musicians & Mics
As the aim was for a live feel, I decided to arrange the players to sit facing each other, with a Sennheiser MKH20 omni mic in the centre — the idea being that this mic would capture a bit of everything, and help me recreate the band's natural balance. In addition, I set up a single close mic per instrument, along with one lead vocal mic which could be moved around to whoever needed it. I ended up using a combination of budget and mid-priced mics for this, with an Audio-Technica AT2020 on the banjo (I've found that the mic has a 'kind' top end which suits instruments such as this), an Audio-Technica AT4050 for the fiddle and mandolin, an sE Electronics Magneto on the acoustic guitar, and one of the new-generation Rode NT1s for the vocals. The acoustic guitar also had a pickup output, which was recorded to a separate track via a DI box.
For the violin, the AT4050 was set up a couple of feet above the instrument, as the sound from a conventionally played violin radiates most upwards and some space is needed between the mic and instrument to allow the sound to mellow a little. Putting the mic up close can produce a very scratchy-sounding result, with excessive bow noise. When used to record the mandolin, the same mic was placed around 300mm in front of the body. While it's normal to avoid the sound holes on a guitar, you can afford to be a bit more lax with mandolins, as their smaller bodies don't produce the same 'body boom' as a guitar, and in any event the player isn't going to stay still enough for you to be too picky.
The acoustic guitar mic was set up at a similar distance but aimed at where the neck joins the body, to avoid picking up too much low end from the soundhole. This usually gives a decent balance between body weight and string definition. I chose the SE Magneto for this role because it has a slightly flattering — but still natural-sounding — character that adds a little 'air' above the 3 to 5 kHz region, where a guitar can sound a bit strident.
The banjo mic was set up around 400mm away, again aimed roughly at the body/neck junction. There's no soundhole on a banjo, of course, as the tensioned skin forms the main radiating surface, and this means that body boom isn't an issue. Most of the sound radiates at right angles to the skin, as the bridge transfers the energy from the string directly into the skin. As with a guitar, moving the mic slightly towards the neck can help pick up a bit more string definition if you need it. I used the AT2020 here as it has a solid mid and upper-mid range but isn't too 'forward' sounding. I didn't want a bright mic that would make the banjo sound harsh. Not getting the mic too close also avoided harshness, though this invariably lead to quite a lot of vocal spill when Peter (the vocalist) was singing. However, as described in the mixing section, this actually helped us achieve a better vocal sound than with the vocal mic alone as it added a nice lower-mid weight to his voice. The banjo is a very loud instrument, and I was concerned initially that it might sound a bit overpowering. Thankfully, though, it actually sat very well with the other instruments for this style of music, requiring only a broad and gentle 1.5kHz EQ lift and some low cut to finesse the sound. I'll bet that's the first time anyone has used the words banjo and finesse in the same sentence!
I spent a little time adjusting the various mic positions for the best results, but there was no point in overdoing the fine-tuning, as I knew that these players would inevitably move around to some extent during the performance.
I wasn't quite sure at first how best to capture the backing vocals, as setting up a separate vocal mic for everyone would have invited too many phase-cancellation problems. With this in mind, I recorded a test run through with everyone playing and singing just as they would do live. I figured that we'd always have the option of resorting to overdubbing the backing vocals if that were necessary but, fortunately, it turned out that the BVs were picked up quite adequately by the various instrument mics, and especially the acoustic guitar mic, as this was the quietest instrument of the three. As we also had a DI from the guitar, that meant I'd be able to juggle the DI/Mic balance to keep the right amount of backing vocals. The output from the central omni mic could also be added in when needed — so, as luck would have it, we seemed to be getting a pretty well-balanced live sound right from the start.
To keep the session running efficiently, I decided that we'd start by recording the tracks on which Andrew (fiddle player and mandolinist) played the mandolin, before switching to those with a violin. Within this schedule, we also tried to do as many consecutive tracks as possible with the same vocalist before switching over, and as a couple of songs required a different banjo tuning we left those until last. There was still a little shuffling back and forth but, on the whole, this approach worked well in that it let us get a lot of songs down in very little time. The band took a couple of breaks during the recording, giving them the opportunity to come and listen to the evidence in the control room. One notable difference between live and recorded music, of course, is that the evidence of any performance flaws remains on a recording for posterity, so you're almost always going to need a couple of takes of some tracks to get a decent performance. Still, nailing the 13 songs required us to record only about 22 takes in all. As we didn't intend to go through any painstaking editing process in search of perfection, we didn't feel the need to use any click tracks. One violin intro proved particularly tricky to get right, though, so in the end Andrew recorded it half a dozen times on his own, before he and I picked the best one and used it to replace the original.
Mixing
Other than Logic's own plug-ins, I used only UAD plug-ins during the mixing stage, simply because I did the final mix on my office laptop hooked up to a UA Apollo interface, which had them included. The interface sent a stereo mix to a pair of Equator Audio monitors with eight-inch drivers. Checking the recordings revealed that although the recording technique meant there was inevitably some spill, this was low enough in level to allow a decent balance to be struck between the close mics, but there was also an adequate amount of the wanted backing vocal spilling into the various mics to create a plausible live-sounding recording. The DI output from the acoustic guitar turned out to be very useful in helping achieve a balance, as in songs where the guitar mic picked up too much vocal it was simply a matter of turning the mic down and the DI up.
I used a little compression on all three instruments (Logic's own compressor on the guitar and fiddle/mandolin and an 1176 on the banjo), each time with a ratio of between 6:1 and 8:1. This just helped smooth the levels to keep them reasonably even. I also used an instance of a Logic bundled compressor with a low ratio (2:1) and low threshold setting on the main vocal, with a UAD 1176 emulation set at 8:1 following this, to act as a two-stage leveller. This two-stage compression setup is actually a fairly widely used technique in which the first compressor reduces the overall dynamic range in a fairly genteel way and the second acts more assertively on any remaining peaks. If I'd been mixing in the studio room I'd probably have gone for the Waves Vocal Rider to do the initial vocal levelling, though this solution worked perfectly well.
The usual sub-80Hz low-cut filter was applied to all the instrument and vocal mic signals to keep out any troublesome low frequencies that didn't belong. As an experiment, I also applied very heavy compression to the omni mic to try to use it as a kind of parallel compression feed, as you might with a room mic on drums, for example. However, that turned out to muddy the sound somewhat, so I reverted to a more modest amount of compression and no EQ, adding just a little of this signal to the mix when needed. Interestingly, Peter's vocals sounded best when the dedicated vocal mic signal was mixed with the nearby banjo mic signal, so to keep the vocal close to the centre of the mix when the banjo was panned slightly to one side, I panned the main vocal mic the other way to compensate.
Level automation was only used in a few of places, just to bolster the odd solo level, but for the most part the players pretty much mixed themselves by playing louder in the solos. I was a little concerned that when recording Andrew singing while playing the violin (alternate phrases rather than both at once!) there may be some audible phase issues, as the violin would be picked up by both the nearby vocal mic and the overhead mic. To address this, I used level automation to pull down the vocal mic by a few decibels when the violin was playing, and to pull down the violin mic when Andrew was singing. Once a balance had been struck, the consensus was that the violin sounded the way it should. Some of the violin parts sounded a little 'edgy' to my ears, compared with how I'd prefer to hear a classically played violin and I did mention this — but I was assured by the band that this was an essential part of the genre and should not be smoothed out!
As the band wanted to make a CD to sell at gigs, some overall mix processing was necessary, including some gentle 2:1 compression and a limiter, which between them trimmed just a couple of decibels off the peaks, added a little density and provided adequate loudness, all without squashing the life out of the sound. I used the UAD Precision Bus Compressor and Limiter for this job. However, I also felt that the sound could do with 'shaping' a little, to get it a bit closer to a live performance vibe, and I felt that this could be achieved by inserting a UAD Ampex stereo analogue tape emulator (prior to the compressor and limiter), which seemed to smooth out and glue together the different elements of the mix nicely. I also used Logic's Space Designer reverb set up on an aux send to add a short auditorium ambience, but treated the violin to a little more reverb than the main vocal, just helping the vocal to stay up front. Even the 1.1 second short ambience sounded a little too wet on some of the songs, and in these cases I shortened this to around 0.7 seconds, using Space Designer's IR envelope controls. This gave a very convincing ambience with no real sense of reverb tail.
The final CD was assembled in Toast, which automatically dithers the 24-bit audio files to 16-bit prior to burning an audio CD. Although fairly basic when it comes to mastering functionality, Toast does allow the balancing of track levels, the fine-tuning of inter-track gaps and, where necessary, crossfades. The final Orange book disc is then suitable for use as a master for Red Book CD production.
Conclusions
Clearly, it's not possible to produce a super-polished recording when working at this speed and with this uniformity of recording setup, but everyone involved seemed to be very happy with both the performance and the end result, which really did end up sounding like three men in a pub enjoying their music! Indeed, it could be said that this genre of music would actually suffer if subjected to the same micro-managing of individual syllables and phase-alignment that is commonplace these days in pop and rock music — sometimes less really can be more.
Audio Examples
Audio files that illustrate the sound captured during this session are available on the SOS web site: http://sosm.ag/jul14-sessionnotesmedia