Chapter 13: Frequency-selective Dynamics
AUDIO FILES (To download all WAV examples at once: 136MB ZIP)
- Equalising Parallel Dynamics Channels: As long as you're wary of phase-cancellation issues, you can create a wide range of frequency-selective dynamics effects simply by equalising parallel dynamics processors. Let's say that you want to add lower-midrange warmth and sustain to this acoustic guitar
(Ex13.01:WAV/MP3 ), for example. A normal parallel compression treatment (Ex13.02:WAV/MP3 ) can add the warmth/sustain attributes that you want, but has also in this case overemphasised high-frequency mechnical noises. Cutting high frequencies out of the compressor return channel focuses the compression effect more into the required frequency range (Ex13.03:WAV/MP3 ). By way of contrast, here's an example of the opposite situation, where a piano part (Ex13.04:WAV/MP3 ) is lacking in high-frequency sustain -- a common problem for pop/rock mix applications. Adding in an HF-heavy parallel compression channel (Ex13.05:WAV/MP3 ) results in a timbre with more high-frequecy sustain (Ex13.06:WAV/MP3 ). Another application of high-frequency parallel compression is adding brightness without overemphasising sibilance. Take this vocal (Ex13.07:WAV/MP3 ), for example, which is already sibilant enough. If you add in an HF-only parallel channel like this (Ex13.08:WAV/MP3 ), it brightens things up without letting the 's' and 't' sounds sand-blast your ear canals (Ex13.09:WAV/MP3 ). Of course, compression isn't the only dynamics process which can be used in a parallel configuration -- gating has its uses too. For example, this kick drum (Ex13.44:WAV/MP3 ) is given controlled presence and weight by mixing in this parallel gate channel (Ex13.45:WAV/MP3 ). Here's the final mixed tone (Ex13.46:WAV/MP3 ), complete with a little of that mix's short blend reverb.
- Side-chain EQ For Full-band Dynamics Processors: The most common applications for this are in gating. For example, here's a snare-drum close-mic recording which captured a lot of spill (Ex13.12:WAV/MP3 ). Spill can often work in your favour when mixing drums, but assuming for the sake of argument that you wanted to use a simple gate processor to remove it, you'd quickly encounter difficulties getting the gate to trigger correctly. When the threshold is set low enough to let through all the wanted snare hits, the kick and hi-hat spill elements also trigger the gate to open sporadically, as in this example (Ex13.13:WAV/MP3 ) -- I've kept the attack and release times fairly fast so that you can easily hear the triggering action. This is where filtering the gate's side-chain can help out: high-pass filtering at 670Hz stops the kick interfering with the gating (Ex13.14:WAV/MP3 ), while a low-pass filter at 4.8kHz protects against the hi-hat (Ex13.15:WAV/MP3 ). It's important to realise, however, that none of this filtering has any effect on the tone of the audio signal passing through the gate, because the filters only affect what the gate's detector circuit is hearing. Equalising the side-chain signal of a compressor can also be useful for adjusting its sensitivity to certain frequencies. One of the most common scenarios for this is when de-essing. For instance, here's a sibilant vocal which requires de-essing (Ex13.16:WAV/MP3 ), and here's what it sounds like after it's been processed with such a de-essing compression setup (Ex13.17:WAV/MP3 ) -- full details of the settings can be found in this Mix Rescue.
- Some Hiss-reduction Strategies: While full-band expansion/gating can deal with noise in the gaps between notes, it's powerless to handle hiss overlaid on the wanted audio, as in this example (Ex13.47:WAV/MP3 ). Fast low-ratio, low-threshold expansion of the high-frequency band can usually achieve around 6dB of hiss reduction without too many undesirable pumping artefacts (Ex13.48:WAV/MP3 ), but you'll need a dedicated spectral expander to take things any further, as in this example (Ex13.49:WAV/MP3 ), which was processed with Voxengo's Redunoise.
- Multi-band Compression At Mixdown: Multi-band compression can help balance different frequency regions of a single sound, and is a surprisingly common requirement for small-studio mixes, especially where the monitoring system's low-frequency response is unreliable. In this Mix Rescue, for instance, the supplied synth part (Ex13.18:WAV/MP3 ) was rather uncontrolled in the sub-150Hz region, and it was only by compressing this section of the spectrum separately with multi-band processing that I was able to confidently increase the power at the low end without headroom difficulties (Ex13.19:WAV/MP3 ). Acoustic guitar parts are also a frequent target for multi-band compression, as a means of controlling high-frequency pick noise. Here's a slightly dodgy-sounding acoustic guitar part which could use a bit of that kind of attention (Ex13.20:WAV/MP3 ). Although you can knock off some of the excess picking with simple fast compression (Ex13.21:WAV/MP3 ), you can improve things even further by aiming this remedy just at high frequencies using a multi-band compressor (Ex13.22:WAV/MP3 ). A similar troubleshooting scheme can also work against overprominent stick noise in drum recordings. These overhead mics (Ex13.23:WAV/MP3 ), for example, have occasional spiky stick noises on the ride cymbals, exacerbated by the heavy compression which has been deliberately applied for artistic effect. Fast, hard compression at the top end smooths off these unwanted corners (Ex13.24:WAV/MP3 ).
- Multi-band Transient Processing At Mixdown: Multi-band transient processing can also deal with guitar pick noise very effectively, as in this example (before transient processing Ex13.25:WAV/MP3 ; after transient processing Ex13.26:WAV/MP3 ). However, the thing I find this kind of processing most useful for is adding low-frequency punch to kick-drum sounds, as in the following examples, the first using Voxengo's TransModder: (before Ex13.27:WAV/MP3 ; after Ex13.28:WAV/MP3 ) and the second using Waves TransX Multi (before Ex13.29:WAV/MP3 ; after Ex13.30:WAV/MP3 ). In both cases you get a lot of extra low-end power, but without any messy low-end rumbling.
- Dynamic EQ At Mixdown: Like multi-band dynamics processing, dynamic EQ has the ability to deal with balance problems between different frequency ranges of the same recording. So, for example, in this drum submix from a live gig (Ex13.31:WAV/MP3 ) the hi-hat was coming through rather too strongly on the snare mic, but gating turned out to be too intrusive a processing approach because the snare mic's spill contribution was important to the sound of the kit as a whole. Expanding the high-frequencies with a dynamic high shelving filter proved to be a more successful approach (Ex13.32:WAV/MP3 ). Where dynamic EQ really comes into its own, though, is where the dynamics of very narrow regions of the spectrum need modification, as in this bass recording (Ex13.33:WAV/MP3 ), where the notes are rendered very uneven-sounding because of the wayward dynamics of the note fundamentals. Compressing the worst offenders with a couple of extremely narrow dynamic EQ bands was just about the only way to smooth things out (Ex13.34:WAV/MP3 ) -- check out this Mix Rescue for full details of this particularly thorny processing problem. The most common application of narrow-band dynamics treatments like this, though, is in dealing with vocal harshness, and here are a couple of different examples of this kind of processing in action. The first is from this Mix Rescue, featuring a female singer: without dynamic EQ (Ex13.35:WAV/MP3 ); with dynamic EQ (Ex13.36:WAV/MP3 ). The second example comes from this Mix Rescue, featuring a male singer: without dynamic EQ (Ex13.37:WAV/MP3 ); with dynamic EQ (Ex13.38:WAV/MP3 ). In both cases, notice how its only on certain more strained notes that the harshness becomes a problem, so it's not something that can be dealt with just using normal static EQ.
- De-essing At Mixdown: Using a de-esser on vocals shouldn't really be rocket science. If you've got a vocal with too much sibilance (Ex13.39:WAV/MP3 ), then use EQ (or the de-esser's built-in 'listen' mode) to find the most shrill frequency, and then dial in just enough gain reduction to tame the problem (Ex13.40:WAV/MP3 ). The main thing to beware of is over-doing the gain-reduction, as this can suck the life out of the vocal and may even make the singer sound like they're lisping (Ex13.41:WAV/MP3 ). Bear in mind, though, that the simple dynamic EQ circuit at the heart of many de-essers can sometimes be useful on other instruments than vocals. The pick noise on this acoustic guitar recording (Ex13.42:WAV/MP3 ), for example, was tackled with a de-esser in the absence of more powerful tools, helping to bring about this final remixed sound (Ex13.43:WAV/MP3 ).
- Affordable Dynamics Plug-ins With Built-in Side-Chain EQ: There are a lot of plug-ins now that have side-chain filtering built in: Cockos's freeware ReaGate and ReaComp; DDMF's affordable NYCompressor; Stillwell Audio's affordable The Rocket; Variety Of Sound's freeware Density32-bit and NastyVCS32-bit; Melda's freeware MCompressor.
- Affordable Multi-band Compression/Expansion Plug-ins: There are now plenty of dedicated multi-band dynamics plug-ins available. As far as freeware is concerned, I'd recommend Cockos ReaXcomp, or for more flexibility have a look at Melda's reasonably priced MMultiBandDynamics, which is a bit of a monster!
- Affordable Multi-band Transient Processor Plug-ins: I've not found anything freeware here, but there are a few fairly affordable options available, such as Melda MMultiBandTransient, Voxengo TransModder and Waves TransX Multi.
- DIY Multi-band Processing: If you're working on a budget, or simply find that available 'one-stop' multi-band processors won't do what you need them to, then you can always implement your own multi-band processing setup using crossovers and multiple full-band processor plug-ins. This is easy to do in Cockos Reaper, for example, using the 3/4/5BandSplitter and 3/4/5BandJoiner algorithms in the ReaJS plug-in in conjunction with Reaper's multi-channel routing capabilities, as explained in this article. If your own DAW doesn't have the same flexibility and crossover functions, you could try implementing the same kind of scheme using Cockos's freeware ReaJS or Robin Schmidt's freeware CrossOver32-bit in conjunction with a separate effects-matrixing plug-in such as DDMF's affordable Metaplugin or Plogue's affordable Bidule. Whichever way you implement it, though, do be wary of the possibility of phase mismatch between the bands if plug-ins with different processing latencies are used.
- Affordable Multi-band Dynamic Noise Reduction Plug-ins: The only freeware multi-band noise-reduction processor I know if is Cockos ReaFIR (using its Subtract and Gate modes). If you need something more flexible for broadband noise, try Voxengo's affordable Redunoise32-bit. If clicks and crackles are also a problem, then the more expensive (but still very reasonably-priced) Izotope RX allows much more user control over the specifics of the processing.
- Affordable Dynamic Equaliser Plug-ins: A couple of freeware options are Platinumears IQ432-bit and Tokyo Dawn's TDR Nova. If you're willing to spend a little money, check out ToneBoosters ridiculously affordable TB Flx, Voxengo GlissEQ, Melda MSpectralDynamics, and (if only because Dave Pensado's a fan!) Waves C1. Not all of these are strictly speaking dynamic equalisers, but the distinctions get a bit blurred with such flexible designs!
- Affordable Dedicated De-esser Plug-ins: While you can achieve de-essing in many different ways with other processors, here are a couple of budget-friendly one-stop plug-ins if you need them: Digital Fishphones's freeware Spitfish32-bit (but beware that it won't report its processing latency correctly to your DAW so it's less suitable for parallel-processing applications), Sleepy-Time DSP's freeware Lisp, and Tone Boosters's extremely affordable TB DeEsser and TB Sibalance.
- Frequency-selective Dynamics In Practice: I use a lot of frequency-selective dynamics processing for salvage purposes in Mix Rescue, but here are some specific examples that are particularly worth looking over. In this article I demonstrated a number of different frequency-selective processing strategies based around equalised parallel dynamics channels. In this article I delved into the practicalities of DIY multi-band processing for a number of the sounds. And this article deals with some in-depth specifics of mastering-style full-mix processing using high-spec multi-band processing.
- More Information On De-essing: Here's a whole article dedicated to the subject of de-essing in its many forms. It includes this downloadable Cockos Reaper project file (2MB ZIP) which demonstrates the mechanics of six different methods of de-essing.
- Source-separation Software Options: Where the mix problems with a recording file prove insurmountable with traditional mix processing, then specialist restoration and sound-extraction software may be the answer. For a good overview of the options and possibilities here, check out Christopher Kissel's monotoSTEREO.info web site. Although the site is officially concerned with upmixing mono mixes to stereo, the technology involved has many mix-salvage rebalancing applications.