Fixing The Mix, Part 2: Advanced Improvements

Last month, Paul White looked at ways to tackle the noise and distortion that can spoil a mix. However, even a perfectly recorded mix can still sound terrible, and this calls for different tactics... This is the last article in a two‑part series.

Sometimes, even properly‑recorded tracks can add up to a less‑than‑sparkling final mix — in the past, I've been asked to remix material that's been well recorded, but somehow still doesn't seem to want to work. In these cases, the problem usually falls into one or more of three main areas: the musical arrangement; the choice of sounds; and, most importantly, the performance itself.

Arrangement

There are no hard and fast rules as to what makes a good arrangement — getting it right is an art and you'll find some people can do it brilliantly, while others can't do it at all. Most of us fall somewhere in between! In pop music, arrangement can usually be sub‑divided into the literal arrangement of the song components — verse, chorus, intro, solo, bridge, and so on — and the instrumentation used for each part. A good arrangement for a traditionally‑structured pop song will be properly paced to create a sense of anticipation before a catchy chorus. The best way to test this aspect of an arrangement is to listen to a demo in front of critical friends. They won't need to say anything — you'll feel yourself squirming at any over‑long links and solos or over‑repetitive sections. Fixing this is usually a case of being ruthless and making every section of the song justify its existence — if it isn't needed, then shorten it or get rid of it altogether.

Mixes can be made to sound louder and less cluttered by gently cutting the mid‑range.

If you're dealing with ready‑recorded material, you might think that changing its arrangement isn't an option, but now that basic hard disk editing is starting to appear as standard with so many sequencers, it is quite feasible to chop up a song into sections and then reassemble them in a different order without the joins showing. I do this all the time using Digidesign's Sound Designer, but you can do the same using something like Steinberg's WaveLab or Cubase Score 3.0 running on a Pentium PC with a soundcard, or with one of the other 'MIDI + Audio' sequencers. Providing you choose your edit points carefully, usually to coincide with a drum beat, it should be possible to make the end result sound as natural as the unedited version.

If you have a hard disk multitrack system, you may be able to transfer the whole song into it from the multitrack master tape; doing this will give you a lot more scope when it comes to moving parts around or replacing sub‑standard parts with sections copied from elsewhere in the song.

Instrumentation

If there's a problem with the instrumentation of a song, you won't be able to completely reorganise this if you're remixing existing material (unless you've been given a clear brief to re‑record parts), but you do have the option to use EQ to reduce the level of parts. You can also mute parts and, in the case of drums, there's sometimes the possibility of replacing sounds. If you have a copy of the sequencer file that accompanied the original recording, you may also be able to replace any MIDI instrument parts with different sounds.

EQ is a useful tool, because it enables us to thin out sounds that are taking up too much space in a mix. For example, an acoustic rhythm guitar used in a busy pop mix might benefit from radical bass cut, so that you end up with a thin sound, almost like a musical hi‑hat in terms of its role in the music. Similarly, if a bass guitar is too boomy, you could try a combination of mid‑range boost and low bass cut to harden up the sound. In many cases, the sense of aural clutter comes from two or more instruments playing at the same time and occupying the same parts of the musical spectrum. You can use EQ to minimise the degree of overlap by, for example, taking some top off a harmonically rich bass sound if it occurs at the same time as a mid‑range pad or lead sound. Similarly, you might be able to take some bottom end off the pad or lead to provide even more separation. This kind of trick is often used to try to separate two similar‑sounding electric guitars, though I have to stress that choosing suitable sounds to start with is invariably best. Using the side‑chain filters on your gate is a very effective way of 'bracketing' a sound to make sure it doesn't stray outside its allocated section of the spectrum.

There are few rules to using EQ, other than to say that if it sounds good, it's OK, but in general, use cut rather than boost where possible, and avoid over‑equalising natural sounds, such as lead vocals, unless you want to create a specific effect.

Performance

Performance errors might include poor timing, poor tuning and even bum notes! There's only a limited amount you can do about any of these problems unless you have a powerful hard disk editing system and you're prepared to spend a lot of time experimenting, but there are a few tricks you can try using samplers and rack effects, that can help fix minor problems.

• TIMING: In a mix where some parts are played just slightly out of time, you may be able to tighten up the bass guitar and drums by using the bass drum to key a gate through which the bass track is being fed. You'll need to set the gate release time by ear, but at the very least, this patch will ensure that the bass guitar only comes in with the bass drum and never comes in before it.

Those with hard disk systems that include a 'quantise audio' facility may be able to reconstruct the offending parts to conform either to the sequencer quantise grid or to a grid taken from another piece of audio that was played correctly, but unless you're just fixing the odd part, this could take a lifetime. If possible, find a similar part elsewhere in the song that is played properly, then copy it and paste it in place of the faulty section — it's a lot quicker. Similarly, if you have a dodgy phrase in a guitar solo, see if you can edit it out to shorten the solo without ruining the musical feel. If you can't shorten the solo, repeating a phrase from elsewhere in the solo may work.

Now that basic hard disk editing is becoming standard with sequencers, it is quite feasible to chop up a song into sections and reassemble them in a different order.

If the drummer's timing is at fault and the drum kit is recorded on several tracks, you may be able to gate the bass drum and snare drum parts, then use these 'cleaned up' sounds to trigger an Alesis D4, DM5 or other trigger‑to‑MIDI‑equipped piece of gear. This is the method often used to replace drum sounds with nicer‑sounding samples — which is something you might want to do anyway — and if you first record the MIDI data into a sequencer locked to tape, you have the opportunity of moving offending beats around to tighten up the timing. If the sound was originally played to a sequencer click track, you'll be able to use the sequencer's own quantise grid, but if it was played 'free', all is not lost — you can still sync up a sequencer, tap in the tempo manually as the song plays, then use the re‑bar facility found in most modern sequencers to automatically create a new tempo map to match your manual 'tap' track. You have to be a pretty accurate tapper to get this dead right, but once you've done it, you can quantise any MIDI data in the knowledge that it will be in time with your song on tape.

Once you've got the relevant drum parts into the sequencer, you can decide whether to trigger samples taken from the original drum recording, or pick a new sampled sound altogether. Whatever you decide, keep in mind that the trigger‑to‑MIDI process takes a small amount of time, as does triggering a sample or drum machine, so you could end up with your triggered beat coming back a few milliseconds later than it should. The subjective delay will probably be very small, but it may be enough to upset the rhythmic feel. In this case, entering a negative track delay facility in your sequencer should enable you to get the timing back to where it should be.

• TUNING: If a track is slightly sharp or flat compared to the rest of the tracks in the song, you can shift it back to pitch in a number of ways. The simplest way is probably to use a pitch‑shifter — even the cheapest models are virtually glitch‑free when performing small shifts, though you may find some general tonal degradation from budget models. Hard disk recording systems offer slightly more scope; Steinberg's Time Bandit, for example, can perform very high‑quality pitch changes, while Logic Audio has a very respectable pitch changer in its Digital Factory section. As a rule, the hard disk, off‑line shifters produce much more natural results than rackmount, real‑time units. Some of the latest 'MIDI + Audio' sequencers have very advanced facilities for pitch manipulation: in Studio Vision Pro, you can convert a monophonic line to MIDI data, edit the MIDI data, and then use the edited data as a template to create a pitch‑changed version of the original. Logic Audio also has some powerful audio‑to‑MIDI and audio manipulation features.

If the part in question is only slightly sharp or flat, you may be able to cheat by adding a pitch‑shifted version to the original, to create a chorus effect. For example, if the track is 10 cents sharp, you could add a ‑20 cents pitch shift, which would create a doubling part 10 cents flat. When the two are played together, you'll hear a chorus effect centered around the correct pitch. Whether you can use this technique depends on the part at fault — strings or pads are usually OK, but other instruments or voices may simply sound wrong with added chorus. The further you have to shift the sound, the stronger the chorus effect will be.

If just the odd note is out of tune, you can use a stand‑alone pitch‑shifter unit with real‑time MIDI control; a synth's pitch‑bend wheel can control the amount of shift. If you can record the control data into a sequencer, you can tweak it until the tuning is spot‑on. Because the tone may suffer when the signal is passed through a pitch‑shifter, you may want to punch in just the offending notes or phrases and re‑record the result onto a spare tape track.

Stereo Mixes

If you're asked to try to sweeten an existing stereo master, the main tools at your disposal are compression and EQ.

• COMPRESSION: Using a compressor on a complete mix will reduce the dynamic range and increase the average energy of a mix, but because this increases the average signal level, you might find that you squeeze all the space out of the mix, which is rarely a good thing. A compressor with auto attack/release settings is easiest to use, but if you only have a conventional compressor, try an attack time of between 10 and 20mS to allow transient sounds to get through. Setting the attack time too fast can result in a dulling of the sound on some models of compressor. Soft‑knee models generally produce the most unobtrusive results, but the subjective difference between individual models seems greater than the difference between soft‑ and hard‑knee variants. Valve compressors may also be used to add an overall warmth to a mix, and many engineers like to use valve processors at the end of a digital recording chain to inject a little warmth into the proceedings.
• EQ: Mixes can be made to sound louder and less cluttered by gently cutting the mid‑range; this simulates the human ear's response at higher sound levels, while also reducing the presence of the mid‑range sounds usually responsible for making a mix sound overcrowded. It's also common to treat a whole mix with an exciter to enhance high‑frequency detail if the mix lacks brightness. I've still found no device more effective than the SPL Vitalizer, which uses a form of EQ to make mixes sound louder and cleaner, but it also has a harmonics control which adds the sonic 'fairy dust' at the top end.

When processing a mix in this way, make sure you audition the result on several different stereo systems, including a car stereo, just to ensure your new mix will travel. I also favour checking mixes on headphones, as these seem much better at revealing low‑level distortions than speakers.

And Finally...

Every mix is different, so there are no absolute rules — only suggestions and hints. What I've presented here are tricks that I use myself and that I know are used by other engineers, but you should still bear in mind that any attempt at salvaging a sub‑standard mix will come second best to re‑recording the offending part.