Studio SOS: Jiang Li

A little Auralex foam put paid to some troublesome wall reflections in Li's room, but Hugh also felt that the speakers were not angled optimally for a seated engineer, so he wedged them up at a slight angle in order to aim the tweeters more appropriately.

The SOS team apply themselves to the task of mixing Chinese traditional instruments with mainstream Western sounds at reader Jiang Li's home studio.

Jiang Li not only records himself and other musicians at his small home studio, but also collaborates with musicians from different parts of the world over the Internet using a system similar to that pioneered by Rocket Networks. However, he was having a few problems with his recording and mixing and so invited us to his studio, which is packed into a fairly small ground-floor room in his Northamptonshire home.

The studio is a little under 3.5m wide and about 2.5m deep, with a ceiling height that is typical of modern houses. The entrance door is in the rear wall of the studio and, once all the gear is in place, the engineer's seat is well under one metre from the door. The monitoring is courtesy of a pair of Mackie HR824s set up on a rigid shelf above the Behringer Eurodesk mixing console.

His latest acquisition was a Mac G5 running Pro Tools LE using a Digi 002 interface, but along with many other G5 owners Li had discovered an unacceptable rhythmic background noise problem when using a Firewire audio interface. This isn't something we could deal with, as it seems like an Apple hardware-related problem affecting some G5 machines — for more details about this, check out Mark Wherry's Apple Notes column this month. Because of this noise problem, Li has kept his Cubase project on his older 450MHz Mac G4, which works perfectly but tends to run up against its processing limits when lots of plug-ins are being used. He also has a fast PC laptop (with an external 200GB Maxtor drive) which is used to run Steinberg's Wavelab for editing, and also Tascam Gigastudio.

Noise Remedies & Monitoring Tweaks

The Digi 002 rackmount interface went dead during the session, but started up fine again after a lunch break. Hugh suspected that it might have been overheating, so he partially removed it from the rack to increase ventilation for the remainder of the session.There was a noticeable hum on the system at high monitoring levels, but because of the way everything was packed into the room we couldn't do any impromptu rewiring. We did however suggest that Li tidy up his mains connections and route them all from the same wall socket, because the overall power requirement was low enough for this not to be a problem. Further improvements can be made when connecting unbalanced sources, such as synths and samplers, to a mixer with balanced line inputs by using the unbalanced-to-balanced cable connection system whereby the balanced end of the cable is wired as normal (using twin-cored screened cable), but at the unbalanced end the screen is left disconnected and the cold (usually black or blue) cable core is connected to the tag where the screen normally goes.

Before listening to Li's mixes in detail, we wanted to evaluate the monitor system's performance, so we auditioned some commercial mixes through the system and felt the result was rather bass heavy. Checking out the switch settings on the rear of the HR824s revealed that Li had set them to the correct half-space mode, but with their full bass extension. I found this setting was too low for my own studio, which is more than twice the size of Li's. I have my bass switches set to their middle positions (47Hz), so suggested Li tried the same. The other issue was that the shelf supporting the monitors was quite high, resulting in the tweeters being well above ear level when Li sat in the chair to mix. Therefore we placed a wedge under the rear of each speaker so that the tweeters were pointing toward where the engineer's head would be when seated.

Repeating the CD playing tests indicated that the balance was now more manageable and, surprisingly, the bass end was fairly even, probably due to the thin rear door acting as a bass trap (or bass leak) and the window in the side wall also tending to do the same. However, the room was not acoustically symmetrical, as the window was to the right and fitted with slatted blinds (helping to break up reflections) while the left-hand wall was bare painted plaster. To help even this up, we brought in some Auralex foam oddments left over from a previous Studio SOS visit and leant one up against the left-hand wall, put one behind the monitors, and propped one up against the hard surface of the door. We left it to Li to find a way to fix these more securely without actually gluing them to the walls — it's easy enough to fix a piece of string across the back and hang them up on picture hooks. A re-test with the commercial CDs showed that the image had clarified to a worthwhile extent, and at this point we felt we had got the monitoring about as right as we could, given the physical constraints of the studio.

Processing Submixes

Once the monitoring system was in order, Paul set about tweaking Li's mix to get the sampled instruments sitting alongside the recorded ones.Next we turned our attention to a mix that was typical of the ones Li was having problems with. This one comprised a mixture of western and Chinese instruments, where the orchestral sounds and some of the percussion sounds were samples, the acoustic bass sound came from a Proteus 2000, and the voice and acoustic Chinese instruments/percussion had been recorded by Li. The problem Hugh and I identified straight away was that the recorded acoustic instruments didn't sit very well alongside the samples — the sampled orchestral sounds had a concert-hall feel, while the acoustic instruments sounded as though they were still in a fairly small room, even though Li had added some reverb.

The vocal also sounded a little thin and uneven in level, and where Li had submixed a number of Chinese instruments to a single stereo track we found that some of the instruments really jumped out while others were getting buried. It seems that the limitations of the G4 had forced Li into working with stereo submixes of sets of the original tracks. It would have been preferable and more flexible had he kept the individual instruments separate and controlled them with fader groups instead. However, this is something he can explore at leisure using his G5 once he's sorted out the noise problems.

To make the best of the existing job, we first compressed the vocal track using Li's Focusrite Penta compressor, which has presets for various applications. Although I don't normally like using presets, the Penta allows further changes to be made once a preset has been selected, so we used the Vocal preset as the obvious starting point and then adjusted the threshold to give five or six decibels of gain reduction on the loudest peaks — which seemed enough to even out the sound. We also added just a small amount of a Chamber reverb patch from Li's Lexicon Alex processor to make the vocal sit in the same kind of space as the sampled backing parts. Good reverb is very important when mixing, but the Lexicon Alex is now rather long in the tooth and probably isn't much of an improvement over run-of-the-mill native reverb plug-ins. Once Li's G5 is behaving properly, he'll have power to spare to run a convolution-based reverb plug-in (such as Audio Ease Altiverb or the new Waves plug-in), which should make a vast improvement in this area.

Rebalancing The Mix

That left us with the Chinese instruments mix to sort out, but compressing this with the dynamics section in Cubase happily made a huge difference. Given the levels Li had recorded, the threshold was set to -20dB and the ratio to 5:1, with the attack time fast and the release time on automatic. These settings worked well to control the loud instruments and to lift out the parts that were previously getting lost.

Before rebalancing the track, we thinned out the string samples by cutting back the low end, and also took off a little top, as they were just starting to sound strident. A further supporting pad was brought down in level. We tried to EQ the string bass sound to get more definition, but found that we needed huge amounts of EQ boost to do this, as there was very little presence in the original sample. The problem with using this amount of EQ was that the string squeak sound Li had added really leapt out, so we used a compromise setting and suggested that Li find some good acoustic bass samples to use for his next project, rather than trying to re-shape the bass sound from his Proteus 2000.

The bright ambience of Li's kitchen was ideal for recording Li's traditional yangqin.To mix the track, we started with all the faders down and then balanced the bass, drums/percussion, and vocal so they worked together. After that it was fairly straightforward to bring in the strings and brass samples, the Chinese instrument mix, and the individual Chinese instruments to achieve a cohesive balance. Where a little reverb was needed to add a sense of space, the same Lexicon Alex Chamber preset was used. When we compared this remix with Li's original version, it transpired that our mix had a higher level of drums, but what was most noticeable was that the individually recorded instruments no longer sounded out of place, and the overall feel of the mix was less crowded because we'd allowed some of the parts, specifically the strings and pad, to play more of a supporting role.

Li asked if we had any special techniques that could be used to create more of a sense of front-to-back perspective, and in some ways we'd already applied these when mixing. As a rule, upfront sounds tend to be less reverberant and brighter than those at the rear, so if you want to keep a vocal at the front, use a shorter, brighter reverb or ambience program, and perhaps use EQ to add a little 'air' into the sound above 10-12kHz using a wide parametric boost. Sounds to the rear can afford to lose a little high end and can be made more reverberant. On this particular track, the vocals were more of a supporting effect than a lead line, so we could afford to use more reverb.

At around this point, the Digi 002 decided to die on us, and after trying to reboot it and checking the connections, we gave up and took a break for lunch — traditional Chinese dim sum presented by Li's wife. It made a welcome change from chocolate biscuits, but don't let that put anyone off offering them on future Studio SOS projects — after all, we have a tradition to uphold!

After lunch, we re-powered the Digi 002 and it sprang back into life, so we wondered if its earlier failure was heat related, even though it was placed at the bottom of the open rack below the Penta, and wasn't exposed to any obvious major heat source. Just in case heat was the problem, though, Hugh temporarily removed the Digi 002 from the rack and we continued to use it this way for a while. However, it suffered a brief relapse later in the afternoon, so we felt it should be returned to base for a checkup, as something was clearly not quite right. Li has since reported that he has rearranged the rack and upgraded his Mac OS to 10.3.1 (which is approved by Digidesign) and the problem seems to have disappeared. Anyhow, it limped on gamely for the rest of our visit, allowing us to experiment with some overall 'mastering' processing. For this part of the process, Li moved our new mix of the song into Pro Tools so that we could use its plug-ins.

Recording Traditional Chinese Instruments

Once we had shown Li our approach to mixing, the conversation turned to the recording of traditional Chinese instruments, of which Li has a fine collection. Although Li had a couple of decent mics, including a nice Audio-Technica 3035 and an ATM31A, he wanted to be able to record up to four musicians at a time and so was in the market for a couple more instrument mics that wouldn't break the budget. I had a couple of SE Electronics SE1s in the car, which provide a decent example of what you might expect from an entry-level small-diaphragm capacitor instrument mic.

Li had been recording instruments in his hallways, which sounded boxy to the ear and gave, I felt, a rather coloured and enclosed tone to some instruments, especially those that couldn't be miked very closely. As Li can use any room on the ground floor of the house for siting the instruments, we checked out the lounge and the kitchen to evaluate their suitability. The kitchen has a hard floor and lots of hard surfaces and so, predictably, sounded quite bright, which we felt could sound good on acoustic guitars and other stringed instruments.

I've heard of playing 'Chopsticks', but this is ridiculous...To confirm this, we miked up a hammer dulcimer (called a yangqin) with a single SE1 mic via Li's Digi 002 preamps and got a very acceptable sound right away. For a 'real' recording however, I suggested that it would be a missed opportunity not to mic this instrument in stereo where possible, so whatever new mics Li bought, he should get two of the same so that he could use them as a stereo pair.

By contrast, the lounge was carpeted and contained a large three-piece suite covered in heavy fabric. The result was a much drier sound that would be ideal for vocal recording. We demonstrated this by me walking between rooms, holding a mic and reading a manual, while Li and Hugh listened (and sniggered!) in the control room. The change in sound when moving from the kitchen to the lounge was massive so, unbeknownst to him, Li already had two very different and very useful recording rooms at his disposal. Even the coloured-sounding hallway could be useful for some sounds, and where good isolation was needed three instruments could be recorded simultaneously in the three spaces, plus a fourth in the studio, with very little spill.

Each Studio SOS visit throws up some common themes and some unexpected questions. In this instance, I think we were able to show Li a slightly different approach to mixing and sound shaping, and exploring the acoustic spaces in his house was also a worthwhile exercise. However, all this is pointless if the monitoring system isn't giving a good account of the truth, which is why we always try to optimise this at the outset. This visit was also unique in that the word duvet was never mentioned! Damn, I've blown it now...