Here are the audio files that accompany February 2019's Mix Rescue article.
The audio files available on this page accompany my Mix Rescue article in SOS February 2019 about helping Michaela Betts (www.michaelabetts.co.uk) improve the mixed sound of a demo recording she did of singer Alessia McDermott (www.belfieldandward.co.uk/actor/alessia-mcdermott). The filenames of these audio examples should be fairly self-explanatory, but here are some additional notes to describe exactly what you're hearing in each case.
This month, Mike helps vocal coach Michaela Betts buff a vocal demo to compete sonically with commercial releases such as Adele's 'Hello'.
Michaela: "What strikes me more than anything is that you have nailed the mix! The piano and vocal can be heard clearly and do not fight, yet I am not struck by there being any compromise in either. The piano is fuller, and the vocal sits in its own space larger than ever. The overriding impression: your mix is smooth and more inviting.
"Overall, I feel very optimistic going forward, because you've helped me see mixing in a new way, and I can see how this will influence my approach to all instruments I have to work with."
The audio is available in a downloadable ZIP file as both 24-bit hi-res WAVs and low-res MP3s, plus these track notes.
Download | 118 MB
This is the raw recording of Alessia that Michaela sent to me for mix advice. It was recorded as an overdub on top of the piano backing track, using a Bock 195 large-diaphragm cardioid condenser mic. As you can hear, there's a good deal of low-frequency rumble on the recording, as well as a strong wind-blast 'thump' at the start of the word "pride" at 0:09.
A 12dB/octave high-pass filter at 145Hz from Cockos Reaper's built-in ReaEQ plug-in did a very respectable job of removing the low-frequency rumble heard in the 01_Vocal_Raw audio file. As a welcome side-effect, the low midrange of the vocal tone was thinned slightly, counteracting some excessive proximity-effect bass boost caused by the close miking position. However, the wind blast at the start of the word "pride" is still overly prominent.
By increasing the high-pass filter's cutoff slope to 24dB/octave, I was able to remove the vestiges of plosive wind blast that remained in the 02_Vocal_HPFAntiRumble audio file without further thinning the vocal tone.
The low midrange levels in the 03_Vocal_HPFAntiPop audio file are a little inconsistent, presumably because the singer was moving around while performing, which caused varying levels of proximity-effect bass boost. To combat this, I inserted an instance of Toneboosters's TB_Flx dynamic EQ plug-in, soft-knee compressing with a 330Hz peaking filter to even this out. The Q value of the filter was 1.5, and the compression attack and release times were both 50ms.
Another area of tonal inconsistency in this vocal recording occurs in the upper midrange region, where strident vocal-track resonances make the timbre a little harsh whenever Alessia moves to her more powerful chest-voice delivery, for example on "how I'm feeling and my" at 0:07. To smooth the tone during these moments, I used the 'Tame' mode of SKnote's affordable SoundBrigade plug-in to restrain some of those resonant peaks in real time. To hear this most clearly, compare this example with the 04_Vocal_DynamicEQ audio file.
To bring out low-level details in the vocal performance, I used Stillwell Audio's Major Tom plug-in as a parallel compressor at a 14:1 ratio, selecting its hard-knee, peak-sensitive, and feed-forward modes for the most assertive compression character. In this example you can hear the effect of mixing this compressed signal in with the uncompressed audio. Note that I've loudness-matched this example with the 05_Vocal_SoundBrigade audio file (using Klangfreund's LUFS Meter plug-in) so you can make more representative comparisons between the two.
The problem with the compressed sound illustrated in the 06_Vocal_ParallelCompression audio file is that the compressor's fast time constants are unduly emphasising the singer's noise consonants. To mitigate this, I applied Toneboosters's full-band TB_DeEsser plug-in to the parallel compressor channel.
Although the parallel compression showcased in the 07_Vocal_ParallelCompressionDeessed audio file was great at bringing out detail and making the vocal level more consistent overall, it wasn't well-suited to managing the levels of occasional overenthusiastic syllables (such as "how I'm" at 0:07). For those, I used an instance of Stillwell Audio's fast-acting The Rocket compressor at its lowest 4:1 ratio, setting the threshold control to catch just the louder words.
Here's the raw piano recording balanced against the processed vocal heard in the 08_Vocal_PeakCompression audio file.
The first thing that struck me about the raw piano recording (see the 09_Piano_Raw audio file) was that it felt rather lightweight at the low end. To remedy this, I used Cockos Reaper's built-in ReaEQ plug-in to apply a healthy 5dB low-frequency shelving boost at 145Hz (Q=0.8).
Both the piano and the vocal featured strong low midrange components in the 200-300Hz region, which made their combined tone feel a little woolly. Using a peaking filter from ReaEQ to cut 3dB from the piano at 330Hz (Q=1.3) helped clarify the mix tonality, as you can hear by comparing this example with the 10_Piano_EQBoost125Hz audio file.
Another more subtle concern of mine was that the piano's slightly hard-edged tone in the upper midrange was causing it to push in front of the vocal in the mix's depth perspective, as you can hear in the 11_Piano_EQCut320Hz audio file. Cutting 3dB at 2.5kHz (Q=0.65) with another of ReaEQ's peaking filters helped to reverse this positioning, tucking the piano more behind the singer, as you'd expect given its accompaniment role.
In order to increase the piano's sense of sustain, I used Reaper's bundled ReaComp compressor with a setting that works well for reducing an instrument's overall dynamic range without significantly changing the internal musical dynamics of the part. The trick is to use reasonably fast attack and release times (3ms and 100ms respectively in this case) in soft-knee mode with a super-low ratio (I chose 1.2:1) and then to reduce the threshold so that there's gain reduction happening almost all the time. You can hear the difference that makes by comparing this example to the 12_Piano_EQCut2500Hz audio file.
In this audio example you can hear the reverb I selected for the piano part. Normally I start from a likely-sounding plug-in preset and do some tweaking from there, but in this case the Lexicon PCM Native Random Hall preset I chose ('Large RHall1') worked perfectly straight away, which is something of a rarity for me!
I used a variety of tactics to keep the vocal reverb from pulling the lead singer backwards in the mix; I deliberately chose a plate reverb algorithm (from Lexicon's PCM Native Plate plug-in), because it sounds less like a natural acoustic; I increased the reverb's predelay time; I EQ'd and de-essed the reverb to make the effect sound appear more distant in comparison to the dry vocal sound. In this audio example you can hear both the piano and vocal reverbs in action within the mix context.
Another treatment I used to bring the vocal closer to the listener was a subtle stereo widener effect based on the classic Harmonizer pitch-shifted delay technique. However, I also used Voxengo's freeware MSED plug-in to remove some of this effects mono component, so as to avoid any phase-related tonal coloration of the core vocal timbre at the centre of the stereo image. If you compare the 15_Effects_VocalReverb audio file with this example, you can hear the subtle difference this makes.
This audio example retains all the processing and effects of the 16_Effects_VocalWidener audio file, but I've also automated both the piano and vocal parts so as to stabilise the balance, increase lyric intelligibility, and draw out the musical emotion for the listener.
Upon referencing the 17_Automation audio file against some commercial releases, the vocal sound felt a little too understated at the high end. Brightening it with EQ caused as many problems as it solved, however, until I'd used Toneboosters TB_Sibalance de-esser plug-in and various instances of region-specific EQ processing to make the high-frequency levels more consistent. The gave the slightly more 'hyped' and chart-oriented sound you can hear in this audio demonstration.