The SOS team tackle an assortment of problems, from computer error messages to boxy‑sounding vocals.
Above: Judy had set up her Mac, mic and speakers in front of a bare wall. Above right: some acoustic foam was all we needed to tame the problematic reflections.
Judy Abraham is a songwriter who records her own material. She plays in most instrumental parts via a keyboard, occasionally getting friends to help out with 'real' instruments. When she contacted us, she'd already established a simple studio system based around a dual‑core Mac Pro running OS 10.4, with plenty of RAM and a dedicated external hard drive for audio. She was using Pro Tools LE, along with a Digidesign 003 audio interface, and had invested in some good plug‑ins, including NI Komplete 5, various instruments from Arturia and Spectrasonics, and Fxpansion's BFD. She also had a copy of Reason, but wasn't sure how best to use it with her other software.
For vocals, Judy had chosen a Blue Bluebird mic "because it looked nice” (not the best of reasons, but fortunately it happened to suit her voice very well!) and had purchased a heavy‑duty SE Electronics mic stand and a Reflexion Filter. An M‑Audio ProKeys 88 weighted USB keyboard acted as the master controller, and a pair of Genelec 8030A speakers perched on the upper shelf of a Quik Lok 600 studio desk took care of monitoring duties. The Digi 003 interface, placed on the desk, doubled as a monitor controller and headphone amp.
Sending Out An SOS
Judy was relatively new to using computers when she contacted us and was still a little wary of them, to the extent that she wasn't entirely confident installing or updating software. She'd been noticing various error messages from the Mac while recording, and because she wasn't very happy with the vocal sound she was achieving.
Judy also had a copy of Logic Pro 8 that she wanted to try — though she lacked the confidence to install it. She'd have preferred to upgrade to Logic Pro 9, but that (and the latest version of Pro Tools LE), required OS 10.5, and she didn't want to risk upsetting the system by updating everything at once — so she'd not yet bought the OS upgrade.
First Things First
After an introductory cup of coffee, Hugh and I decided to start by looking at the computer. To find out more about the problems, we asked Judy to load up one of her songs and then opened the system performance meter, to view the disk and CPU activity meters. Although the average CPU load was under 50 percent, occasional spikes came very close to pushing the system over the edge, and on busier songs that seemed to be the probable cause of the error messages. There were also some disk-activity spikes, though nothing excessive. Providing the peak plug‑in load was kept within the CPU's limits, the system seemed perfectly stable, so we outlined to Judy the reasons for the computer problems, along with some strategies for avoiding them.
Like many users, Judy was unclear as to what made most difference in terms of performance: the CPU speed, the amount of RAM installed and the hard drive all contribute. Users suffering CPU overloads often assume that more RAM will fix the problem, but that's unlikely: an adequate amount of RAM is necessary for good performance but doesn't make the CPU run any faster. Once you have more than three or four GB of RAM, the only real benefit of increasing beyond this is if you use lots of large samples in software instruments that can make use of the extra RAM.
Hard-drive speed is largely unrelated to CPU speed, but can still be a bottleneck. The more tracks you play at once, the faster the hard drive has to work to manage the data flow. Its job is more difficult if your tracks are made up of lots of tiny regions, too, because the hard-drive head has to locate each piece of data before the computer can play them back. A simple way to avoid this is to consolidate fragmented track segments into a single audio file, as the drive head then only has to find the start of the file once.
Hard drive activity may also spike when many audio tracks start at exactly the same time — as they often do at the beginning of a chorus, for example. For this reason, I try to ensure that the parts on each track have slightly different start times, even if that means they contain a little silence. For example, if the drums don't play on the first couple of bars, I might trim the start of that track so that it begins a little later. We tried these strategies on the test song and both the CPU and disk-activity meter calmed down.
Latency
Paul examines Judy's recording setup, including her Blue Bluebird mic: Judy was fortunate in having picked a mic whose tonality suited her voice — but the recordings themselves sounded very boxy.
Judy had noticed an irritating echo when recording vocals, but had tried to live with it. We found that the buffer size was set at 1024 samples, which is ideal when mixing but causes a high latency, which is off‑putting during overdubbing if you monitor via the computer when recording audio and playing virtual instruments. We suggested that when she's tracking, she uses the smallest buffer size that will work (128 samples is a practical value in most cases), but increase it to 1024 when mixing, to help smooth out CPU spikes if adding CPU‑hungry plug‑ins such as reverb.
We tried to demonstrate the reduction in latency with a smaller buffer, but after resetting it to 128 samples, we could still hear a delay when playing MIDI keyboard parts. Switching to the Mac's on-board audio interface reduced the latency to the point where it was pretty much unnoticeable, so a certain amount of head-scratching followed. I suspected that the problem was related to the 003 box or its driver software. Ultimately, we restarted the computer and the problem simply went away, so we had to put it down to 'just one of those things': sometimes problems demand a restart, and it doesn't usually mean the system is about to have a meltdown!
Software instruments can tend to be CPU hogs, and Judy had some 'big guns' Spectrasonics stuff on the system, as well as FXpansion's BFD2, which she uses for drums. She didn't seem to have made the connection between inserting a lot of plug‑ins and the system putting up CPU-overload error messages — but soon recognised the pattern when we explained the connection! Where computing power is limited, the simplest way to avoid problems is to convert software instrument tracks to audio tracks once you're happy with them, but keep the MIDI data tracks in case you need to go back and make changes. Most DAWs now include a neat 'freeze' function that does this automatically. If you need to make changes you simply 'unfreeze' the track, tweak, and re‑freeze.
Vocals & Effects
Above: the mic stand was incorrectly set up, with the central pole touching the floor — so it wasn't isolating the mic from footsteps and other vibrations from the floor. The correct configuration is shown in the second pic, below, with the pole raised off the ground.
Judy's lack of success at setting up reverb sends in Pro Tools had led her to use multiple reverb plug‑ins as insert effects on individual channels. Even though Pro Tools' D‑Verb is relatively CPU efficient, using half a dozen or more instances of it at a time inevitably drains CPU resources, so we went through the process of setting up aux sends and buses in Pro Tools. This allows you to send some of the signal from any track to the bus track. If the software is used in that way, you can put a single reverb on the bus to be shared by different channels. This worked fine, but we had to remind Judy that to hear the solo'd vocal with reverb she'd have to solo the aux return (or make it 'solo‑safe') as well as the main vocal track.
Despite the fact that she was using a Reflexion Filter, some of Judy's vocals had a 'boxy' quality, and on at least one recording there was an undesirable flange-like effect. The delay may have come from forgetting to mute the speakers when recording, as the mic position was close to the computer desk — and with the original buffer size of 1024 samples the delay would have been very noticeable — but we couldn't figure out how the flange effect had been created, unless it was a result of Judy and/or the mic moving relative to the speakers while singing.
Looking at the mic setup, the first problem to resolve was that the mic stand had its legs splayed out horizontally, which left the centre of the mic stand touching the floor. Not only did this make the stand wobble, it also circumvented any shock‑isolating effect of the rubber feet, and made the mic more susceptible to mechanical vibration. Simply readjusting the leg collar raised the centre column away from the floor, improved the stand's stability considerably, and cured the mechanical noise issue.
We also moved the mic position slightly, so that it was roughly level with the 'mouth' of the Reflexion Filter (it had previously been a couple of inches further away). Hugh was keen to rebuild the Reflexion Filter hardware to improve its physical balance, but as the heavy SE stand was perfectly stable, we decided that we should leave it well alone on this occasion.
Backs Against The Wall!
Judy was recording vocals with her back to a bare wall, and we suspected, given the boxy room sound, that sound was reflecting back from this, over her shoulder and into the front of the mic — because that's the one angle the Reflexion Filter and most similar devices won't be able to cover. We decided to hang some simple Auralex Foam panels on the wall behind her, and to continue these around the corner onto the side of a fitted cupboard. She could then record facing the Reflexion Filter, with her back to an absorbent corner.
We put up four panels in all, so that wall reflections would be minimised both from the singing position and from the mixing seat. Thin (two‑inch) foam like this is only effective at mid-range and high frequencies, but it can make a huge subjective difference to recorded vocals, and also to clarity when monitoring. As the flat was rented, we couldn't glue foam to the walls, so we glued an old CD to the back of each one near the top edge and then used the hole in the CD to hang over a map pin, which was pushed into the wall. We got Judy to make a test recording of her vocals and she was amazed by how much cleaner the result sounded.
Vocal Processing
Judy also wanted to know how best to treat vocals in a mix. She'd been adding a bit of delay and some default‑setting D‑Verb, but with a decay time of about four seconds, it all sounded a bit cave‑like! We adjusted D‑Verb to produce a fairly short plate reverb, with a bit of pre‑delay, and saved this as a user setting that Judy could select in future projects.
We then explained how to configure a compressor to level out the dynamics of Judy's vocals and make them sit better in a mix. Again, we saved a preset, but explained that as the ideal compression setting depends on the level of the actual vocal track, you still need to adjust the threshold every time to get the right amount of gain reduction. In Judy's case the vocal sounded good set to around 4dB of gain reduction. Hugh was also keen to point out that she could afford to record at a lower level, as she'd been trying to get close to the top of the meter when singing, which risked clipping if she sang slightly louder when doing the actual take. Indeed, she told us she often had to redo recordings because of unexpected clipping. Leaving around 10dB or so of headroom makes life much easier — both when tracking and when mixing, and there's no noise floor problem with modern 24‑bit audio interfaces.
Going Soft
Affixing an unwanted CD to the back of a foam panel makes it easy to hang on a single pin — which is particularly useful in rented accommodation.
Paul White hangs foam in the corner, to help reduce the boxiness of recordings, and to provide a suitable space for vocal recording.
The key to making Judy's setup work for recording was to have sufficient treatment behind the recording/listening position to cut out unwanted relefections when monitoring, and to complement the Reflexion Filter when recording.
Over lunch, we set Logic Pro 8 installing, but to save time we omitted much of the optional content, which Judy might not need. Once installed, Logic booted up without problems, taking a few minutes to verify Judy's third‑party AU plug‑ins and stopping only when it found Spectrasonics' Trilogy — which was not compatible with the system. (Judy had updated to Trillian after first discovering this but hadn't uninstalled Trilogy).
As Judy was new to Logic, we created a default song template, with screensets for the Arrange page, the Mixer page, and one page with just a transport bar at the bottom, so Logic could be controlled while displaying Reason on the rest of the page. We decided on eight mono audio tracks and eight software instrument tracks as a good starting point, and added three aux sends to the mixer, the first two feeding two different Space Designer reverbs (a vocal plate and a small concert hall for the instruments) and the third feeding a delay. A further aux channel was set up, with Reason acting as its source via ReWire. Though she hadn't really used Logic before, Judy really liked the way it worked and was particularly impressed by the selection of built‑in plug‑ins.
When working in Pro Tools, Judy had tried a light version of Celemony's Melodyne pitch and time‑correction processor for fixing problems. Melodyne is a very powerful piece of software, and probably the best one available, but Judy found it quite hard work, as it requires a lot of user input, and wanted something simpler to use. In Logic, we tried the bundled Pitch Correction plug‑in, which works by setting a scale of notes and a speed of correction — much like Antares' original Auto-Tune plug‑in, which it imitates. This worked very well with her voice, as her pitching is generally pretty good anyway, and the plug‑in just added the final polish. I also set up a vocal compressor, and we made another test vocal recording to show off the vocal processing and to fine-tune the plate reverb. Again, Judy was really impressed by how solid and clear the vocals sounded in comparison with her earlier recordings.
We spent half an hour going over the basics of using Logic, focusing on the essentials such as setting recording sources, instantiating plug‑ins and basic cut, copy and paste editing. We also looked at the fundamentals of automation. Judy seemed very keen on the way the program works, and, given that she makes extensive use of software instruments, she felt that in some situations it would be a more useful platform for her composition work than Pro Tools.
Finishing Touches
While we were putting the finishing touches to our 'Logic 101' session, Hugh repositioned the Genelec monitors to get them aiming correctly at the engineer's head, and fussed with the cabling in his usual, capable manner! By late afternoon, Judy felt more confident about getting the best performance from her computer, and was particularly pleased that we'd come up with a method for achieving a good vocal sound that suited her voice. With that, we thanked her for lunch and moved onto our next challenge — London's North Circular road during rush hour!
Reader Reaction
SOS reader Judy Abrahams.Judy: "The sound treatment you put up works so well; I never would have thought it could make that much difference when recording vocals, but it really does: I get much cleaner vocals now.
"You answered most of my questions, and I'm using my setup with more confidence than before. Creatively, I don't have to keep starting and stopping because of some technical glitch or hitch, loss of sound and so on. I'm so glad you took the trouble to sort my system out.
"I don't seem to have any problems with losing sound, or crashes, or anything so far. The only thing I did was reinstall the rest of Logic's sound library, as there were a lot of instruments without sound! It was good that you set up the template, especially for assigning reverbs and delays. I'm also now using compression, and that seems to be working well for me. Thank you!”