You are here

Mix Rescue: Obscuresounds | Media

Mixing & Recording Advice By Mike Senior
Published July 2010

Get more from July 2010's Mix Rescue by listening to the accompanying audio file examples.

The following audio files accompany my mix rescue of Lawrence Eldridge's track, 'Infernal Machine' (/sos/jul10/articles/mixrescue_0710.htm).


SOS Mix Rescue: July 2010 issue by Sound On Sound

Download all Hi-Res WAVs | 211 MB

BussCompOn

Here's a section of the final remix with master buss compression from URS Console Strip Pro. The emulated analogue compressor selected was the classic Neve 2254 in Limiter mode with fast attack and release settings. In order to achieve even gain pumping on kick and snare beats, a side-chain high-pass filter was inserted at 40Hz, and the compressor's Mix control was set to 51 percent to reduce the impact of the limiting on the drum transients, and in this configuration gain-reduction peaks of up to about 10dB didn't produce unacceptable side-effects.

BussCompOff

This file is identical to BussCompOff, except that the master buss compressor has been bypassed. Compare this with side-by-side with BussCompOn in your DAW to compare the differences most easily.

BassUnprocessed

Lawrence's original bass part was programmed in Ableton live using a sample from the 8-bit PCM sound chip of an old Commodore Amiga personal computer.

BassFinal

This is the processed bass part as it appears in the final remix. Initial spectral balancing was provided by Reaper's ReaEQ plug-in (high-pass filtering at 20Hz, -2dB low shelving at 510Hz, and a 5dB peak at 1.6kHz with a Q value of 2) after which an instance of the ReaXcomp multiband compressor squashed the sub-150Hz region using the default 2:1 ratio, 15ms attack, and 150ms release settings to give gain reduction in the region of 5dB. Roughly 8.5dB of combined make-up gain in that band gave an overall increase in the low bass power, but with creater control on account of the compression. Following this, a little further sculpting with Stillwell Audio's Vibe EQ peaking bands comprised 3dB boost at 820Hz, and 1dB cut at 315Hz. There was only one send effect used on bass: a short ambience reverb patch from the Sonnox Oxford Reverb plug-in.

BassNoMulticomp

Here is the same processed bass sound as in the BassFinal file, but with the multi-band compression bypassed so that you can hear what it contributed to the overall sound. Line up this file against BassFinal in your sequencer for the most revealing comparisons.

OverheadsUnprocessed

This is a section of the stereo drum overhead track during the song's first chorus section. It already sounds nice and natural without processing, but in this case natural wasn't what Lawrence was after, as you can hear from his own mix in the OriginalMix file.

OverheadsFinal

This is the same section of overheads track heard in the OverheadsProcessed file, but processed as it appears in the final mix. A combination of transient smoothing (from Reaper's own Jesusonic Transient Controller), limiting (from Tin Brooke Tales TLS Mastering Maximzer), and multi-band compression (from Reaper's ReaXcomp) were used to smooth away the peaks, and then a layer of rhythmic level automation was added to simulate a pumping compression effect. On its own the result sounds almost ridiculously over-processed, but when heard within the context of the track it helps give the drums a dynamic movement that suits the otherworldly lo-fi feel of the song as a whole.

OverheadsNoMulticomp

The ReaXcomp multi-band compression setting used for the OverheadsFinal example was specifically designed to eradicate some over-zingy stick noise on the ride-cymbal hits. Just the top end of the spectrum above 5kHz was processed, but that was limited with an instantaneous attack time, 6ms release, and a 100:1 ratio. With quite a low threshold this regularly knocked 10dB from the high-frequency peaks, and firmly scotched the problem. This audio example demonstrates what the tracks would have sounded like without this multi-band processing, so you can compare it with the OverheadsFinal file to gauge how much of an impact it had.

SnareFinal

Here's the final processed sound of the snare as it appears in the remix. It's made up of two layers: a snare sample from an old Commodore Amiga's 8-bit sound chip, and a simulated live snare track (complete with kit spill) from XLN Audio's Addictive Drums virtual instrument. The sample was simply boosted 4dB with a 350Hz low shelf EQ from U-He's Uhbik-Q plug-in. The 'live' track, on the other hand was processed first with Tin Brooke Tales TLS Mastering Maximizer to even out the levels some of the snare hits, then with GVST's GClip in soft saturation mode to soften the peaks further and add some grime. ReaComp was used to duck the Addictive Drums snare whenever the sample was playing. Two EQ plug-ins were added after that: U-He's Uhbik-Q equalizer shelved off around 8dB above 5kHz, while a narrow peaking cut in Reaper's ReaEQ targeted an unwanted pitched resonance. Send effects amounted to a little bit of global ambience from Sonnox Oxford Reverb for blend, a little mono reverse reverb effect to bulk out the snare sample's sustain, and a larger dose of gated reverb from Stillwell Audio's Verbiage to give density and width.

SnareNoSCDyn

In order to reduce the level of the Addictive Drums snare track during sampled-snare hits, I used Reaper's ReaComp on the former track, but fed its detection side-chain from the latter. The compressor itself was then set up with fast 3ms attack and 8ms release times and an infinity:1 ratio, and then the threshold was set to duck the snare hits by more than 15dB. To avoid any bits of attack transient breaking through the ducking, I also dialled in 7ms of lookahead into the detector path. This audio example shows how the layered snare would have sounded without this ducking compressor, and if you compare it to the SnareFinal file you can hear that it's more open in the midrange — not usually a problem, but in this case not what I felt the song called for.

SnareNoGatedReverb

While I wanted to give the snare sound something of the width and depth that Lawrence had achieved in his original mix using a long room effect, I was keen to avoid a long reverb tail. The solution was to use a short gated reverb from Stillwell Audio's Verbiage. Because the reverb tail was cut off abruptly, I was freely able to adjust Verbiage's reverberation algorithm parameters for the desired width and texture without any risk of clutteing the track over time. This audio example shows how the layered snare would have sounded without the gated reverb patch. If you break out in a rash at the mere thought of using gated reverb, then compare this audio file with the SnareFinal file to see how this effect can actually work well when used subtly. There's still life in it despite the excessed of the '80s!

LVFinal

Here's a section of the final processed lead vocal track. Processing plug-ins used included Reaper's ReaEQ, Universal Audio's 4K Channel Strip, and Universal Audio's Helios Type 69 for EQ; and Universal Audio's LA3A and Stillwell Audio's Rocket compressors, the latter with 12dB of 8kHz boost in its side-chain to suppress sibilance. As far as send effects are concerned, I used a stereo ADT patch and a three-sixteenth ping-pong delay from Reaper's own plug-ins; short ambience and longer chamber reverbs from two instances of Sonnox Oxford Reverb; and Universal Audio's emulation of Roland's Dimension D stereo processor, working in its fourth mode. However, a good proportion of the audible effects actually came directly from Lawrence's original mix, a trick achieved by layering his effect vocal stem alongside my own processed and effected raw vocal.

LVFinalNoSCDyn

As I mentioned in the notes accompanying the LVFinal file, many of the effects on the vocal actually came from the stereo vocal stem of Lawrence's own final mix. However, the practical problem I had was that I wanted less of his dry vocal and more of the effects, and the way I achieved this was to duck Lawrence's vocal stem quickly and severely whenever my own processed raw vocal track was active. Had I not done this, it would have sounded like this current audio example, in which the vocal doesn't come across nearly as clearly as in the LVFinal example.

LVFinalNoFXTrack

Just to give you an idea of how much Lawrence's own processed and effected vocal stem contributed to the final remixed vocal sound, for this audio example I've muted that vocal stem. Compare it with LVFinal — it's a substantial difference!

SFXAdditions

This file shows how I laid out the different added ambience, sound-effects, and textural samples throughout the song to aid the arrangement dynamics and reduce the need for reverb to fill out the production. If you line this file up against the main Remix file, you can see how the sound effects placement relates to the song's different musical sections.

RemixNoSampleSFX

This is a version of my remix but without any of the added samples I spotlighted in the SFXAdditions example. A/B this with the main Remix file to get an idea of how much difference the samples made to the end result.

RemixNoSendEffects

This is a version of my remix without any of the reverb or delay send effects, again to demonstrate the extent of the difference these made to the production. Compare to teh main Remix file in the first instance, but also compare with RemixNoSampleSFX to evaluate the relative impact of the samples relative to that of the reverb/delay effects.

OriginalMix

This is the original mix of Obscuresounds' song 'Infernal Machine', as it was sent in to Mix Rescue by SOS reader Lawrence Eldridge.

Remix

This is the final remix created using Lawrence's original multitrack files, with a little help in terms of textures and atmospherics from a few media sound-effect libraries..    

Published July 2010