We consider some of the pros and cons of mixing in surround, and hear how experienced surround producers take advantage of the format, and tackle the problems.
In last month's instalment of this series, we looked at some of the difficulties that need to be overcome when you try to set up a surround-capable studio using project studio equipment. We also spoke to two surround recording engineers with very different setups to see how they were addressing these problems, both of whom we'll hear from again this month: Rik Ede, a Dolby employee who runs his own business from his project studio composing 5.1 surround soundtracks for computer games, and Peter Cobbin, Senior Recording Engineer at Abbey Road Studios.
So let's move on from putting together your system, and look at how you can make use of surround when you have a suitable system up and running. Obviously, mixing for surround is a more complex process than for stereo, with a much greater range of options for the spatial placement of instruments in a mix, and of course there's the '0.1' channel — the subwoofer — to consider as well.
One of the constant refrains from the professionals working in surround who we've spoken to so far is that they have a feeling of greater freedom when mixing in the format, which they attribute to the current lack of clearly defined guidelines or accepted conventions for surround mixing. However, one effect of this is that engineers differ over what's acceptable in surround production and what isn't (even more than in stereo production!), and this is clearly evident in the interviews with Rik and Peter this month, where they frequently contradict one another. Of course, neither of them is correct or incorrect; they just have different approaches.
So how should we approach surround production? My own feeling on this is that the answer should be 'cautiously'. We've all seen badly set-up consumer hi-fi systems, so consider the implications of a badly adjusted surround playback system, where the scope for incorrect placement or misuse of the speakers is so much greater, because there are so many more of them. Most people buy surround systems for home cinema use, and all such systems have a built-in facility to help the user match the levels of the speakers, usually by routing bursts of noise to each speaker in succession. Nevertheless, it's my guess that even where the user has bothered to do this, there are still likely to be discrepancies of several dBs between the Centre and front Left/front Right speakers, and between the front three and the Surround speakers. And that's without considering the sub-bass speaker, which is probably running 10dB too hot just to make the earthquakes and space battles on video DVDs sound impressive (or is that just my system?). Imagine what damage that could do to a carefully crafted music mix!
Now, it's probably fair to say that as more surround music becomes available, people will cotton on to the fact that they have to set up their systems properly. But until that happens, should we construct our surround mixes so that they take account of potential problems with the playback system (and if so, what do we assume people will have done wrong)? Or should we just assume that everyone is playing back 5.1 on perfectly calibrated systems and leave it up to end users to sort things out?
Another factor some people like to take into account when they mix is the DVD player being used to listen to finished 5.1 epics. Some DVD players have only stereo speakers attached, and when they are given a 5.1 surround mix to play back, they 'collapse' the 5.1 mix into a stereo-compatible affair that will play back without losing channels or musical information. The technical term in use to describe this process is 'fold-down' or 'downmixing', as explained in Part 5 of this series (SOS December 2001), and many engineers like to create cautious 5.1 mixes that produce reasonable-sounding stereo when downmixed.
There's also the question of whether you should consider not only the playback system, but also the storage medium containing your 5.1 mix. If your surround mix is destined to end up on a DVD‑Video disc with video material of some sort, it will have to be data-compressed in order that it can fit on the disc along with the video data, as even DVDs only have so much space and playback bandwidth, and video is very data intensive (see the 'Balancing The Bit Budget' box later in this article). As explained in Part 4 of this series (see SOS November 2000), the most widely used data-compression formats for DVD‑Video are DTS and AC3 (Dolby Digital), and these can also affect your audio. Should attempts be made to minimise the effects of these compression algorithms during mixing, as in the days of mixing for Dolby's Pro Logic format (see Part 2)?
Opinions vary tremendously on all these topics. Some contend that the only completely safe way to do a surround mix is to use just the front Left and Right speakers to produce a conventional stereo mix, and ignore the Centre and Sub speakers, because some people don't have them connected or are driving them at the wrong level. This is the same school of thought that advocates conservative use of both Surround channels, so that a stereo downmix still comes out sounding listenable.
However, many people feel that this careful approach rather wastes surround's potential, and that it's much more interesting to place important sounds in the dedicated Centre channel, so that they cut through a mix that bit more effectively, rather than relying on the 'phantom image' generated by a conventional left/right stereo speaker system (for more on the difference between the 'phantom' and 'real' centre in a 5.1 mix, see Part 5). Technically, this latter view is perfectly valid, but I would nevertheless urge caution, as it's also true that the most important musical components tend to be in the centre of the mix, specifically vocals and solo instruments. Think back to your last mix, and remember how hard you tried to get the perfect vocal balance by making adjustments of less than half a dB. Now imagine the same mix played back on a surround system with all the vocals mixed to the Centre channel. If the Centre speaker hasn't been set up properly by the end user, the vocals will sound noticeably too loud or too quiet. My feeling on how to deal with this problem is that it would be prudent to hedge one's bets, and combine vocals routed to the Centre channel with a phantom-centre image generated in the front Left and Right speakers, which will be unaffected by an incorrectly configured Centre speaker. Incidentally, this was also the approach to use of the Centre channel taken during the preparation of the 5.1 mix of the Super Furry Animals' album Rings Around The World (see Part 6).
Rik Ede is a firm believer in unashamed use of the Centre channel; "I very rarely have things in the Centre and front Left/Right speakers at the same time, and I like to keep things like the snare drum, lead guitar and vocals pinned to the Centre speaker." However, Rik also pays careful attention to how his mixes sound when played back on a variety of systems, from surround-compatible Soundblaster and small multimedia speaker-based PC systems, right up to full-range surround monitoring setups.
At Abbey Road, Peter Cobbin has yet another approach. He's a definite fan of the Centre channel, but doesn't use it for everything he wants to be located centrally. "There's a lot of discussion about whether we need the Centre speaker. My experience is that classical music often works well without it, but I couldn't do pop music without it — it really nails the centre solidly. However, phantom and real centres sound very different, and both work for some things but not for others, so I find it important to try both as I'm building a mix. We've had a modification done to the SSL J‑series desk here at Abbey Road to allow me to do that quickly. Originally, the desk could be switched globally to operate with either a Left-Centre-Right panning mode, with a dedicated Centre channel, or a straight stereo Left-Right mode, with a phantom centre. The individual Pan knobs on each channel have all now been fitted with a switch element, so that if you pull the knob it pops up, like some of the other rotary controls on the desk, and that allows individual channels to be switched between Left-Centre-Right and Left-Right panning."
If there's much debate about the role of the Centre channel, the situation is even worse when you come to consider the subwoofer, originally designed to handle low-frequency effects for cinema. Should you leave its upper roll-off set at the Dolby standard of 120Hz, or should you reduce this frequency so that anything added to the subwoofer channel sits beneath the normal range of full-range speakers? For example, setting the LFE frequency at 80Hz, as THX recommend, would minimise the overlap between the full-range surround speakers and the sub. Even then, can you afford to put more than just a hint of energy into the Sub channel, in case the end user has it set too loud?
Again, people have differing beliefs on this subject, as we'll hear in a moment, but for me, pragmatism wins the day; I think it's wise to avoid feeding anything into the 0.1 channel that you can't afford to lose. In other words, make sure the mix sounds OK without it, then use it to add that bottom octave to deep bass, kick drums or sound effects. It's important not to feel pressured into using the sub for the sake of it, or because you think that omitting it will make your mix will sound thin on one of those compact domestic surround systems that uses tiny satellite speakers in conjunction with a subwoofer. These usually have their own bass management systems that feed low frequencies from the main channels into the sub so as to produce a nominal full-range response, and this happens whether you make use of the 0.1 channel or not.
This time both Rik Ede and Peter Cobbin are in agreement on basic use of the Sub channel, although they still use it in different ways. Rik: "If you don't have a subwoofer connected to your playback system, then depending on how your system is set up, information destined for the Sub channel may be redistributed to the rest of the speakers, but not necessarily equally. For that reason, it's important not to feed any essential information to the sub, because it may not get reproduced on some domestic systems. To check everything is OK, I always run through finished tracks with my sub switched off and check that they still sound OK."
Another concern of Rik's is the upper roll-off frequency for the sub. "If you go back to when Dolby Digital was invented in 1991, the sub roll-off was set at 120Hz, and that was fine for film use, but they didn't envisage dance music producers coming along and wanting to use it. Around four years ago, Dolby themselves were recommending that those mixing music should steer clear of the sub altogether, but once you start creating music for 5.1, as opposed to just mixing existing music in surround, the sub's extra low-frequency control is useful. I roll mine off at 66Hz so that what is coming through the sub is really only the very low-frequency, gut-moving stuff, and it falls below the range of a typical full-range system. Pianos will cross over that 66Hz point at the low end, but I don't mix any of that into the sub because it can tend to make everything a bit 'washy'. With bass guitars, I might send a little to the sub, but not much. Percussive bass synths can sound good with some low end fed to the sub, though. It's a personal thing, but I like using it just to reinforce, say, the bottom end of kick drums and things like that, so that they really hit you in the chest.
"My 02R mixers don't handle the filtering for the sub, so I produce my Sub channel at full bandwidth, then apply low-pass filtering at 66Hz to the material that feeds it, via a plug-in with a fairly steep rolloff — 18dB per octave usually does it."
At Abbey Road, Peter Cobbin has a full-range five-way B&W 801N monitoring system, and doesn't bother altering the roll-off of the sub, leaving it at the suggested Dolby standard of 120Hz. However, he is just as cautious as Rik Ede when it comes to using the 0.1 channel. "The end user may not be using the Sub channel, so I tend to spread bass over the front three speakers, and only use the sub to create extra depth for special effects. Sometimes I'll run a little of the low end of the drum kit to the sub, but always after I have made sure the basic five-channel mix works by itself, without the sub switched on at all. This is really why I like to mix on a full-range system, not one with satellite speakers and a bass-management facility."
We looked at Abbey Road's facilities for 5.1 Surround mixing last month — this month it's the turn of the mastering department. The studios have an in-house DVD-authoring department (Abbey Road Interactive), who put finished AC3 or DTS audio files together with edited video and create the graphic front-end needed for finished DVD‑Video releases, but before this stage can happen, the surround mixes need to be properly mastered and converted to AC3 or DTS format in the first place. This task falls to the engineers in Abbey Road's Audio Restoration suite, who are also responsible for creating high-resolution MLP-format audio from the surround masters for use with DVD‑Audio discs (for more on MLP — Meridian Lossless Packing — see Part 4 of this series). As their title suggests, the Audio Restoration engineers also clean up old mono and stereo audio recordings, and are responsible when required for upmixing this restored audio to create 5.1-compatible mixes.
Completed six-channel surround mixes are sent from Studio 3 on eight-channel Tascam DA98HR tapes, or placed as six-channel digital files on Abbey Road's central computer server, where the Audio Restoration engineers can access them. The six-channel mixes are then loaded into the studio's mastering systems. At the time of our visit, both SonicStudio HD and SADiE digital audio workstations were in use for mastering, but the Audio Restoration suite had moved over to SADiE. Mastered files are passed through Dolby AC3 or DTS encoders, and metadata (see part five of this series) is also set up for the files at this point via the workstation being used. The encoded audio files are then passed on to the Interactive department for DVD authoring.