Studio SOS: Monitoring Woes

Because the monitors were set up along the long wall of David's narrow studio room, the bare wall directly behind the mixing position was causing monitoring problems. A duvet slung between mic stands over this area created an immediate improvement in the sound, so it was suggested that David mount a rail on the wall to allow a duvet to be hung there more permanently.

David Ashman felt that his mixes lacked energy and were sounding 'too digital', so the SOS team set off to his home in Bristol to sort out his monitoring system and mix processing.

David Ashman called us in because he was having trouble getting any energy into his mixes, and he also described them as sounding 'too digital'. However, he also thought his monitoring environment might be giving him problems, so we set out for his Bristol home studio to see if we could help. As David was a regular reader of the Studio SOS series, he was aware of the unwritten rules, and so had coffee, chocolate digestive biscuits and chocolate cake prepared in readiness!

David's studio is now centred around Emagic's Logic Audio running on a rather noisy dual-processor Mac G4 and using an Emagic EMI 2|6 USB interface, though his previous musical experience had been based on Sonic Foundry's Acid running on a PC (which he still has). The EMI interface was powered up via USB, and we had to boot up the computer twice before we got any audio. In theory USB power should be OK, but as the interface is also designed to be able to work from an external PSU, I felt that getting a suitable PSU and powering it from the mains might be safer.

Dealing With The Room

The bedroom housing the studio is long and narrow, with the speakers (passive Spirit Absolute 2s) set up on stands along the longer wall and spaced rather too widely for the listening distance. The surfaces of the room were completely untreated and painted bright green, so that when Hugh was taking photographs I felt almost as though I was on a green-screen Star Wars set!

Once the rear-wall reflections had been tamed, critical listening to David's mixes highlighted the ineffectiveness of his monitoring system at low frequencies, a deficiency which had led him to make unsuitable choices of instrumentation and made mixing very hit and miss.

We always start out by listening to some original mixes in the room prior to making any changes and suggestions, and it immediately became evident that there was more than one problem, the most troubling being the inaccuracy of the monitoring system. The choice of sounds used in the mix was also questionable, though without accurate monitoring, choosing the best sounds is very difficult. We've come across Spirit Absolute 2s in this series before, and again we found that they were simply not telling us what was going on in the critical lower octaves, though, being passive, the choice of amplifier used to drive them would also have had an effect.

Checking David's mixes with Hugh's rather nice Sony MDR7509 headphones confirmed that there was quite a lot going on at the low end that we simply weren't hearing. There was no practical way to fix this using the Absolute 2s so we recommended David invest in a full-range pair of active monitors. We also suggested that he move them closer together, possibly by moving both his Mac and PC systems into a single computer desk rather than having the speakers separated by two computer desks. David also had to do a lot of unplugging to play back mixes, so in the absence of a hardware mixer, a monitor controller such as the Samson C*Control would be ideal, as this also handles the level control for active monitors, provides multiple source switching and includes headphone monitoring and talkback.

Because the wall behind the mixing chair was completely bare, we tried the old duvet trick to help soak up some of the reflected sound, with a view to tightening up the stereo image and reducing coloration. As luck would have it, the rear wall was lightweight plasterboard on a studding frame, which meant it would absorb or allow through a lot of the low end, leaving the duvet only to soak up the mid-range and high frequencies. This meant the duvet was more effective than if the wall had been solid brick. Having proved the principle, David said he'd fix up a rail from which he could hang a duvet more permanently. Using a rail to hang the duvet a few inches from the wall is more effective than pinning it directly to the wall.

Because the side walls were a fair distance from the monitors, and because one was largely taken up by a window, we didn't advocate applying acoustic foam to them. Instead we suggested that the bare wall opposite the window (to the left of the monitoring position) be used to house a bookshelf or something similar, just to break up side-to-side reflections. That was about as much as we could do on the monitoring front, though I invited David along to my studio so that he could compare some different monitors, including my Mackie HR824s and whatever else was in for review at the time.

Here you can see the layout of David's studio, showing the overly wide spacing of the monitor speakers with regard to the listening position, which was compromising the stereo imaging. A narrower desk setup for the two computer workstations would allow the monitors to be placed closer together. Also note the suggested positioning of the double duvet (using a wall-mounted rail to provide a little clearance between the duvet and the wall) and the recommended addition of some bookshelves at the left-hand side to scatter sound reflections. Diagram: Tom Flint.

Problems With Over-processing

Getting back to the mixes, we thought the bass sounds were rather bland and lacking in punch, so we asked David to open up the original Logic song to see what he'd done. It turned out that David had not yet rigged up a MIDI interface to his new Logic system, so all his parts had been written in step time, giving them a somewhat mechanical feel. Most of the sound sources came from samples loaded into the EXS24 software sampler and, rather than optimising the track levels, David had inserted a limiter plug-in into each track to prevent the signals peaking. This worked quite well on the drum tracks, but was unnecessary on the other parts. Having said that, limiters can be useful on resonant synth parts, as some of those resonant peaks can be very loud. He'd also used Logic's DJ EQ plug-in on most tracks, along with compression, but again these proved largely unnecessary. Where EQ is needed, the new Channel EQ is more flexible and sounds good.

For reverb, David had used four instances of Audio Ease Altiverb set up on four different send busses. These took up a very significant proportion of the available CPU resources, and we felt the use of reverb was rather heavier than the track needed, which robbed it of some of its impact. As an experiment, I used a single Logic Platinumverb set to a short, bright, plate-like setting. It may not have been as refined or realistic as the Altiverb, but it sounded more appropriate in this instance, and the CPU load dropped to almost nil. David had also bought a second-hand TC Powercore PCI card, but had not yet installed it, so I suggested that he try the excellent Classicverb, which is bundled with the card, when he got it running.

David had also used the default EXS24 sine wave for one of the parts, which is fair enough, but he'd also added EQ and compression. MIDI parts can often be controlled using Logic's MIDI dynamics rather than compression, though compression can be effective on some drum or percussive sounds. However, using EQ on a sine wave is not very useful, as EQ can only change the level of a sine wave, not its harmonic content, because it has none! David had learned the theory behind this, but had left the EQ switched in by default. Instead, to make the sine sound more interesting, we tried the Logic Phase Distortion plug-in, which turned out to be remarkably effective and gave the track an organ-like quality with a hint of FM edge. A tempo-related Tape Delay plug-in was added to a sequenced piano sample to give it more of an arpeggio feel.

Bass & Drum Tweaks

As the bass parts were the obvious weak points, we isolated them and listened to the two layered parts David had created in isolation. I suggested that one of the parts could be replaced using the Logic ESM monosynth plug-in, as it is capable of very TB303-like bass sounds that I felt might work well in the track. Judging the effectiveness of this new bass sound was very difficult over the Absolute 2 monitors, which were again telling us there was no low end, though Hugh's headphones confirmed that there was plenty of bass! For the other bass part, which used a synth sample, we again processed it using a moderate amount of phase distortion, just to dirty it up a little.

Passing one of a pair of layered drum parts through Logic's EVOC20 vocoder, and modulating it between two different filter-bank settings, helped to add interest to an otherwise repetitive part.

David had also layered two drum parts, one of which had been treated with a flanger. We tried Logic's EVOC20 filter bank, which we activated in time-limited demo mode, in place of the flanger and used a tempo-locked LFO to morph between two filter settings to create a fluid, rhythmic sound that really emphasised the higher percussive parts. The movement of the modulated filter added interest to what was otherwise a very straightforward drum part.

David had also dosed his drum parts with a generous helping of Altiverb, but, because the parts were not split over several tracks, all the drums got the same treatment. This invariably makes the kick drum too reverberant, so the ideal solution is to split the drum part over two or more tracks, with the kick on its own track. However, it is possible find a suitable reverb to work on a whole drum part, provided that you roll off the low end of the reverb sound. You can do this using the filtering parameters in the reverb itself, or by inserting a low-cut (high-pass) filter before the reverb, set to between 150Hz and 200Hz. This filtering, combined with a short reverb of between 0.6s and 1s, can work well when you have no way to split up the drum part, and in any event you only need enough reverb to take the dry edge off the sampled drum sounds (in this case GM drum kit samples).

Getting Rid Of The Mechanical Feel

At a purely artistic level, I felt that writing everything in step time using Logic's Matrix Edit window and then copying blocks gave the drum part a somewhat sterile feel, so I added a few hi-hat beats (again in the Matrix Edit) which varied from bar to bar over a four-bar cycle, just to demonstrate that small variations can make a part sound more organic. David admitted that programming good drum parts with the right groove was quite difficult, so in addition to the more obvious drum loop sample CDs, I suggested he try some of the Keyfax Twiddly Bits MIDI drum loops, as they are inexpensive and can be used to trigger any drum sounds. David liked this idea, because he was worried that if he used sampled loops, these would be recognisable.

We also listened to some of the more creative loop-based music that David had made in Acid, much of it incorporating sounds he'd recorded, often via a simple binaural mic setup. Of course the spatial magic of binaural recordings only works properly over headphones, but the overall effect was powerful and imaginative. Because David was obviously comfortable working this way, we felt he might try creating some rhythmic parts in Acid using the PC, then save these as stereo WAV files to be imported into Logic, where they could be chopped up, copied and looped to be used as the basis for new compositions, rather than relying on off-the-shelf loops. I'd also have liked to have heard some tracks with David's guitar more in evidence (which he records via a Line 6 Pod Pro) as just a track or two of a 'real' instrument can add a lot of life and depth to an otherwise all-MIDI composition.

Another processing technique we experimented with was using Logic's Tremolo plug-in as a chopper/panner by setting the wave shape to square and setting the two outputs to opposite phases of modulation. In other words, when the left side was turned up, the right side was turned down, and vice versa. By using the tempo sync function to chop 16 times every bar, interesting rhythmic modulations can be created that work well on pad and even vocal parts — although if you were to use this on vocals, it would be best confined to short passages.

Rap Vocal Processing

Next we turned our attention to a rap vocal part that David had recorded using a young vocalist who'd actually performed in the same room as the computer, and monitored over speakers rather than headphones. Consequently there was some spill, but as the vocalist had a fairly strong voice, this didn't cause too many problems. To even up the level, I used Logic's Compressor plug-in with a ratio of 6:1 and a hard knee, adjusting the threshold to give around 6-8dB of gain reduction on the loudest phrases. The attack time was set as fast as possible, with about a quarter of a second of release. The rather overbearing Altiverb was swapped for a short, bright Logic Platinumverb to create a more intimate, 'in your face' sound, though, given time, I'm sure that we could have called up an Altiverb ambience program that would have done the job just as well, and with a better sense of real space.

David had actually recorded the vocal directly via a Digitech Vocalist Workstation vocal processor, but as he also had a very competent TLA D1 valve mic preamp lying around doing nothing, we felt he might get better results if he recorded using that, adding any necessary vocal processing while mixing. In fact I think it's a good general rule to record as much as possible with no EQ and no processing, at least until you gain more experience, as you can always go back to a clean slate if the mix is proving difficult.

Mastering In Software

David's final question related to mastering, as he wasn't sure whether to buy dedicated hardware or to rely on plug-ins. After David first contacted SOS, Reviews Editor Mike Senior had processed one of his earlier recordings using a Drawmer DC2476, which opened up the sound and revealed high-end detail that had been lost in the original mix. Again the monitors didn't tell us how well this worked at the low end, so we double-checked using Hugh's headphones.

Because David didn't want to pay out for hardware if he could get an acceptable result using software, (and we'd already added new monitors and headphones to his shopping list!), I tried the simple approach of putting Logic's Multipressor multi-band compressor in the main stereo mix insert point, followed by the Limiter. I set the compressor to work over three bands, rather than the default four bands (just to make it more manageable) and set the crossover points at around 120Hz and 5kHz, so as to keep the critical mid-range intact. The compression ratio in each band was set to around 1.2:1 and then the threshold was adjusted to give just a few decibels of gain reduction in each band. On most mixes, this means having as threshold setting of between -30 and -40dB. Using a slightly higher ratio on the bass band can increase the low-end density.

Here you can see how Logic's Multipressor plug-in was set up to subtly enhance David's mixes. The number of bands was reduced to three, and low-threshold, low-ratio compression was applied in each band. In addition, the make-up gain for the middle band was reduced by half a decibel to create a slight 'smile curve'.

This approach to compression provides a homogenising effect by compressing gently over a wide dynamic range. This is different to the way individual tracks tend to be treated, where it is normal to set the threshold much higher and then use a higher ratio so that only excessively loud peaks get stamped on. The tonality of the mix can be balanced by altering the relative levels of the three compressor bands, so if you want the 'smile curve' loudness effect, all you need do is pull down the mid-band by a couple of decibels relative to the high and low ends. The limiter takes care of any signal peaks, and the threshold should be set so the gain reduction is no more than around 4-5dB on the signal peaks. I have to admit that Logic's Limiter puzzles me, because when you set it up so that you'd think it was limiting everything at around 0dB, the output peaks at around -5dB, and it doesn't have an output target control where you can determine the exact value at which limiting occurs. For my own work, I always use the Waves L1 when I need limiting, which I still find to be the most useful and intuitive of all the limiter plug-ins I've tried, though I know there are some equally well-specified contenders from the likes of TC Electronic.

Tackling A Noisy Computer

Paul suggested reducing the noise from David's PC by placing thick acoustic foam on the rear wall to absorb sound emerging from the open back of the cabinet, as well as using thinner foam on the inside surfaces of the cabinet. Diagram: Tom Flint.Before wrapping up, we decided to have a go at quietening the extraordinarily noisy 'mirror door' Mac G4, which was sitting in the bottom of a lightweight, open-backed computer cabinet with the door open. Closing the doors didn't help much, as most of the noise seemed to be coming out of the back, so we moved the cabinet away from the wall slightly and redeployed the duvet behind it just to see how much of a difference it made. The improvement wasn't dramatic, but it was noticeable, and now closing the doors did make an improvement. Without wanting to specify anything too elaborate, we settled on suggesting acoustic foam on the wall behind the cabinet, with thinner foam used inside the cabinet and door to kill reflections inside the box. This should improve the noise situation by a few welcome decibels, while still allowing air to circulate around the computer for cooling.

For critical vocal parts, David could run a mic cable out onto the landing, where the computer was to all intents and purposes inaudible. Although the landing was long and tended to colour the sound, experiments with the ubiquitous duvet confirmed that hanging one across the landing to isolate the end where the recording was being made improved the situation considerably, though hanging another duvet over the wall and door at the end of the landing that the singer would be facing made the sound cleaner still.

Recommendations

Yet again we found that what was originally described as one problem had numerous facets, the most serious of which was the choice of monitors. We've come across Absolute 2s twice in the course of our SOS visits, and though they are fine in the mid-range and at the high end, we've found them to be completely inadequate at the bass end, leaving the user guessing as to what's going on down there. They are fine as secondary monitors to tell you how something might sound on a small domestic hi-fi, but I can't recommend them as main monitors. Consequently, replacing these with something more honest should be a first priority, and improving the monitor geometry by reducing the spacing between the speakers should also be high on the list.

The amount of room treatment needed is pretty minimal, and a heavy double duvet across the back wall should fix the most serious problems, as the partition wall behind it will work as an impromptu bass trap. I must stress that the duvet trick is just a cheap and cheerful fix, and doesn't give you the same results as a properly designed studio, but from a pragmatic viewpoint such simple fixes are usually enough to give you an acoustic environment you can work in without the sound being too misleading.

As David is a keyboard player (as well as a guitarist), we feel he should get a MIDI interface up and running as soon as possible, so that he can use real-time feel and expression, rather than adding notes in step time. Drum parts can be livened up using sampled loops, MIDI drum grooves, or parts he creates for himself in Acid, and once the monitoring has been sorted out it should be much easier to choose sounds that work properly in the mix, rather than relying on processing to try to re-shape things later.