Mixing & Mastering Engineer: Karl Junga

Karl has enjoyed UK, US & International chart success with #1s as a mixing & mastering engineer, producer & as a signed artist. An uncommon skillset.

Biography:  Mastering The Art Of Audio

A classically trained keyboardist, with a specialist Mathematical Physics BSc and 3 years PhD Artificial Intelligence research at a leading UK university.   Tested to the limit as a pro studio runner and time-served as a freelance audio enginee in high-end analogue and digital as a mixing engineer and then mastering engineer. Signed  to both Major and Indie deals as an artist. With decades of practice-led research at the leading edge of music technology as practitioner and associate lecturer. In short, a thoroughly rounded critical insight across  the art and the science.

He started piano lessons at seven after going nuts about a Bontempi organ ad on TV and ultimately walked away from a lucrative career in computer science to work in a studio as a runner, and never looked back. “Music is a calling, I just got drawn to the things that interested me the most. Ultimately, I felt I had to go with that. The maths and AI was great, but it didn’t grab me 360 like music.”

For a complete CV see the Linkedin icon revealed when clicking the floating contact button below right.

A Mixing & Mastering Engineer With Sonic Vision

Fast forward 35yrs and, it turns out music, maths, physics and computer science are forever intertwined. Not to mention psychology, wordsmithing, business, marketing and some seriously heaving-lifting soft people skills. Karl is clear that computers and AI are just tools; that music needs soul and vision. “It takes real musical connection to read what a mix or master needs. From old-school 50s sounding rock n roll buss compression to contrasting processing across the stems that will bring out the punk, the grimy beats, the drill edginess, or enhance a beautiful, modern open sound that sparkle and shimmers. There are so many ways to finesse a mix and a master sonically. It’s the final creative process, not just a technical ‘polishing’ process – especially now with an emerging pool of mastering engineers who can mix just as well as they can master. Combine those skills with full projects or stems that now remain ‘live’ till the final delivered master, and it’s opened a whole world of possibilities that the ‘legacy’ mastering engineers just can’t help with or advise on. “

What Does A Mix Or Mastering Engineer Even Do?

“But most people have no idea what a good mix or mastering engineer brings to the listening experience. If we get it right no one outside of our peers should notice. The public should just notice the great track and artist. If people are saying it’s a great mix or master, it’s either because the song’s not great, or occasionally the song is class and the mix and master are very spacious and ‘shiny’. It’s rare to get ‘tech’ accolade when the track is popping out the speakers 3D due to choice mix techniques and harmonic distortion. That’s all attributed to the band or the ‘raw’ recording. So it’s the nature of the work. It would take me days and days of solid work to go through Discogs and gather the credits never given by artist or label. But equally, possibly the key quality one needs in these roles is to get over yourself. It’s not about you, it’s about the music. The music needs to be out there – doing its miraculous thing of helping people laugh, cry, get together, breakup, meditate, heal, set the scene for the game ,sell stuff, bussup the club, wake up the neighbours… grieve for your loved ones. It’s a fundamental fabric of human life.”

Joining The Sirkus

Karl has an impressive analogue output dating back to the early 90s. He co-founded Sirkus Studios in 1997 when he saw the big studio exodus that the nascent digital recording revolution would inevitably bring. He’s been the Managing Director since foundation. Not many studios survive 25+ years and counting, so that truly sustainable business nouse is also worth noting. Sirkus was digital from the get go, Soundscape and Creator on Atari with analogue outboard and digital mixer, then fully in the Emagic Logic box by around 2000. And since then the long list of vinyl and streaming masters alike has been delivered using Sirkus’ superb analogue emulation and high-end passive monitoring .

Less Is Truly More… Until It’s Just Less

He learned early in his career that the idea of a mixing or mastering engineer just ‘correcting’ everything or ‘separating’ everything was nonsense. He thinks it may be a hangover from the days of engineers in white labcoats and producers in tweed suits. Perpetuated down to a host of ill considered myths like the one about frequency ‘redundancy’ – those charts of harmonic range of violins etc that help students follow their non commercially trained educators and “lop off the bottom and top frequencies all too eagerly to ‘clean it up’. “I had one 3rd yr degree student just couldn’t understand why instruments refused to gel in their mix. I just grabbed all the faders and bypassed the high and low cuts on every track with 2 clicks. His face was a picture when the mix suddenly grew a pair. Low cuts are for badly recorded tracks with rumble. High cuts are just plain daft on virtual instruments. “

Seeing how top engineers he shadowed went about a mix, and listening to the soloed tracks of their finished mixes made it very clear to him that many elements of a great mix sound odd on their own. Great mixes are not assembled by pushing up faders of great sounding individual stems. “it’s obvious really” he says, “we don’t hear an orchestra like a set of a hundred instruments all perfectly miked. And we have the age old painters’ problem of how to replicate the experience of being at the performance. The adrenaline, the crowd dynamic, now that’s a real tough ask.”

Psychoacoustic Smoke & Mirrors

So he says knowing how to construct a mix becomes an impressionistic process: ” ..we have to attempt to recreate the visceral impact of truly immersive complex music performed at potentially loud volume, … on tiny speakers! Getting a record over the line fully mixed & mastered is more like making a film than the trite idea of ‘balancing sound to clarify everything’. The way one picks out elements or juxtaposes them is like cutting, zooming and focussing. Mastering is akin to ‘colour grading’ sonics into the right era or vibe. So arrangement segments become ‘scenes’ which assemble into the whole experience of the record. Our brain has finite capacity to simultaneously focus on who’s doing what and why. So we have to direct the listener to what’s important for the narrative and what mixture of feelings is relevant; and then differentiate that from the supporting players and scenery. Many of the cues as to how to interpret the spectrum of melody, harmony, poetry and sonics are psychoacoustic. Like the meme of putting lullabies over the scariest movie clips; but incredibly subtle subliminal signalling that our brain registers even if we don’t consciously understand what we’re hearing. “

“A good example of this psychoacoustic power is that our brain just won’t assemble a sensible picture of a cleanly recorded acoustic flute mixed way louder than a rock guitar stack. It will try to put the flautist right by your ear and get confused. It’s preprogrammed, distortion means ‘loud’ – as much as darkness and no echoes coming from the ground in front means FREEZE! Pure survival – don’t walk off the top of the stairs or the cliff. We don’t need to understand those subconscious reactions, they’re hard-wired.. But assembling them back in when they’re absent takes knowledge and practice. Get the psychoacoustic cues right and our brain may just trigger some of the same dopamine etc pathways even if we’re just listening on air pods or a mono radio. Psychoacoustics is a hugely important area, look at archeoacoustics now showing us that ancient temples and spaces were clearly designed to enhance sound to enhance spiritual experience or maybe healing.”

Analogue Heresy

His opinion is that analogue is better… and digital is the best way to achieve that analogue sound! Read that again. “Differentiating each track through the signature harmonics of legendary analogue processing is modern digital magic, not second best to the ‘real’ kit. Neve strips sound great. Add vocals through a Summit chain or EMI console strip, and drums through an API strip. Then keep all their harmonic signatures without imposing the Neve or SSL summing engine over everything. Things really start to pop out from each other.”

“In reality, mixing on a real analogue console just homogenises all the tracks.” Karl mixed on the best analogue consoles for a decade and is not keen on going back there either as a mixing or mastering engineer. But caveats that there may be a strong production reason to go old-school – for the decisions it forces. The creative imperative…and maybe just revelling in the imperfections and the sheer knob twiddling geekery.”

Standing Out From Crowd – Do You Need A Good Mixing &/Or Mastering Engineer?

Shown in a 2019 small sample listening study to substantially outperform LANDR masters,  Karl also outscored the other three very well known US/UK audio mastering engineers.† Names withheld as the poll is extremely tentative in the scheme of things. But he offers insight into the potential end-stops of AI in creative endeavours, based on his PhD AI research at Nottingham University and subsequent study: “LANDR is pretty smart, but it won’t outperform really experienced audio mastering engineers any time soon. It’s ‘trained’ to sound like an  ‘average’ good master. Not make a track stand out within that genre. It’s an inherently limiting behaviour till we get actual intelligence that can throw the right sonic curveballs.”

“That needs a deep connection with what the entities within a mix may be doing and why. Intuitive artistic connections remain impenetrable to current machine learning or knowledge engineering. (The first step of automating – mapping the process.) Bounded but complex goals like winning a chess match are a relative breeze for current computing power. No super Grand Master can beat a supercomputer in unrestricted competition. Because it just brute-force models the possible outcomes beyond even genius ability within the allowable parameters. Take those ‘rules’ away and creative processes are pretty much infinitely unbounded. They are not addressable by big data munching, neural net learning or language models that mimic what we say and how we say it. Those models like Grok can discuss this in great depth – disarmingly well. But they can’t do it well. They can make arresting images – but they are arresting, not high art. They can resynthesize John Lennon’s voice, but not his inspiration or creative process. I doubt even AI recoding its own code without supervision unlocks that door, because it’s about consciousness itself for me. Something the reductionists argue is the result of chemical pathways in the brain. But I’m in the consciousness is separate camp. Either way, as I see it, generating this sort of ‘Actual’ Intelligence’ as opposed to ‘Apparent’ Intelligence’ is more or less the same problem as replicating human experience and consciousness. Tbh, regarding creativity, Douglas Adams’ eons of Deep Thought ’42’ is as close as Elon Musk et al may get.”

“That’s not to say that Elon’s concern at the potential threat to humanity of current paths of AI development is not real. AI doesn’t need to be truly intelligent to wreak decision making havoc – or help solve the biggest problems we face. But skills that involve intuition and empathy are a long way from AI’s scope. Averaging some smart sonics from proven releases is impressive, for sure. LANDR beats novices all day long, and it’s a useful tool for musicians finding their way,. But its output is the exact opposite in some ways of what experienced mastering engineers want or do.”

” I’ve tried it, and as well as that inherent ‘safe and average’ problem, there seems to be lot of multi band activity which messes with phase coherence. Also, trained human hearing is still qualitatively leagues above the best FFT etc quantitative audio analysis tools. So even fairly simple subjective stuff like realising there’s lots of odd-order harmonics embedded in a mix already, so we may need to look at even-order processing or linear phase, is not accessible to machine learning. AI would first have to be able to hear with the perception of an expert human, and then understand why this record conjures up echoes of Nirvana or Drake or whatever… and know what type of mastering processors or processing might push that further… and why that might be a good idea with this new artist… as opposed to a bad idea with this established artist. Go code that for me in Python please GPT.”

Sound Research

Along the way, Karl has designed and installed studios large and small – right up to educational complexees. Initially fully analogue setups, latterly favouring Dante or AVB over ethernet for simultaneous tracking from any room to any room. “I know how poor the resultant signal to noise ratio gets from the miles of even well shielded cables and Krone boxes. Pristine takes degrade down every connection. The real problem is the phenomenal complexity of the interconnected local audio and mains wiring biome. I want the magical harmonic shimmer of the circuit – not the intractable impedance/ inductance/ earthing/ RF crud. This is only achievable with digital – it’s also massively more cost effective on initial outlay and maintenance.”

More Signal Less (Unwanted) Noise

That’s why he goes for great emulations over real outboard. Combined with Ableton, Cubase, FL Studio, Logic Pro X, ProTools, Reaper or Wavelab Pro, depending on the job and which DAW the project was produced in. “Sometimes it’s better to stem out at 32bit into a DAW that’s more suited to mix buss structures, even though we have about 1500 of the very best plugins available. But ultimately, we’re looking to take the vision of the producer and get the max s/n ratio and therefore highest bit depth summing all the way down the mix chain to the final master, then enhance that vision with choice mix attention where it really counts… and master that vision ‘live’ on the buss where we get total 64 bit precision right into the final wavs, FLACs whatever.”

“With the headroom of most DAWs it’s better gain structure than premastering out and then mastering. It’s not hard to find a mastering engineer who’ll disagree with this, but most mastering engineers are not trained mix engineers. Tbh, the harder bit by far once you can master proficiently is getting the mix to be top class. Go ahead and try it for yourself with just an Ozone Maximizer or LoudMax to get the right commercial LUFSi in-place on your mix/master buss. Render that, line it up and hot solo-swap a/b it against a rendered premaster mix you’ve mastered in another project to the same LUFSi. The benefit is very, very clear on good monitoring systems and cans.”

How To Compete With The Best Mixes & Masters

On a final note Karl asserts that many ‘top’ PCM lossless masters out there will clip their subsequent mp3s. To sound like the best on myriad streaming services, radio, big club systems etc, masters should be fully compliant with Apple Digital Masters , and optimised for high bit depth or low bandwidth delivery alike … and vinyl cut ready. He’s against the idea of masters per platform other than perhaps vinyl, citing a potential nightmare of confusion at distribution. Though he also laments that mastering is ultimately a compromise: “.. if everyone had a full range studio setup, we would all choose much more dynamic masters. Knowing that no one is ever going to hear it exactly like it is in the studio is a hard fact of life for a pragmatic purist. Bob Katz’ lovely idea of calibrated listening for the general public, and Dolby’s wizzbang multichannel repeat of the promise of quadraphonics are the antithesis of bluetooth bars and one speaker behind the chair reality. ‘Real’ listening environments are the statistical norm not the exception. “

“People mostly just don’t care about fidelity or their Spotify normalise loudness option, but mastering engineers have to make permanent decisions about delivering sound as good as it possibly can be, wherever it gets played. The desire for high dynamic range lossless is real, but the audience is TINY. The fact is that good mastering engineers apply world class limiting that is far better than random proprietary maximiser algorithm applied by the streaming service. I’m sure that’s why the vast majority of Billboard 100 tracks are done by a real mastering engineer and at -7,8,9,10 LUFSi even though Spotify specifies -14.”

† Blind listening test results from a 1st class De Montfort University final year BSc Music Technology dissertation: 64% selected KJAMM master, next best engineer selected by 52%. Each of the 4 named mastering engineers completed a different track, blind to the LANDR master. Approx 50 test completions. Author, Will Russel BSc.