You are here

Mix Rescue: Gypsy Jazz

Jérémy Dutheil & Solal Poux Trio By Sam Inglis
Published May 2021

The upper photo, from the recording session, shows Jérémy (left), Mathieu (centre) and Solal, and their respective mic placements. I’ve also captured a still from a YouTube video which was one of Jérémy’s mix references, showing the Coles ribbon mic on lead guitar, apparently paired with a small‑diaphragm capacitor mic.The upper photo, from the recording session, shows Jérémy (left), Mathieu (centre) and Solal, and their respective mic placements. I’ve also captured a still from a YouTube video which was one of Jérémy’s mix references, showing the Coles ribbon mic on lead guitar, apparently paired with a small‑diaphragm capacitor mic.

We try to improve a gypsy jazz mix, and discover that even the best references need to be taken with a pinch of salt.

Using reference tracks is a good thing. It helps to stop over‑familiarity and ear fatigue leading you down wrong paths, especially if you’re new to recording and mixing. But referencing also has its pitfalls, as this month’s Mix Rescue illustrates.

Jérémy Dutheil is a fine accordion player who jointly leads an excellent gypsy jazz group with guitarist Solal Poux. His recording, made in a large living room, captured the group in trio mode with second guitarist Mathieu Chatelain. When the time came to mix, Jérémy had identified a couple of appropriate commercial tracks to use as references — but no matter what he did, his mixes refused to sound like them! Having posted several iterations of his own mix on the SOS Forum, he then sent the source tracks to me to see what a fresh pair of ears could contribute.

Small Is Beautiful

Live recording is normal in this genre, and it was a very refreshing change to work on a project containing only four mono audio tracks, none of them requiring any timing correction, tuning or other performance fixes. The accordion had been tracked using a pair of Prodipe STC‑3D large‑diaphragm capacitor mics, while each guitar had enjoyed the attentions of a Prodipe A1 pencil mic. Everything was in tune, nothing was clipping or noisy, and the playing was a treat.

However, it was also clear why Jérémy had had difficulty making these tracks sound like his references. One reason was that although the references were in the same genre, they featured different instrumentation and material. Low‑frequency content in Jérémy’s track was limited to whatever the accordion and guitars could generate, but both of his references had a double bass. And whereas both of his references were mid‑tempo affairs, the Charleston that was causing Jérémy problems was a fast dance tune played with more energy.

It also seemed likely that the reference tracks had been recorded differently. There was no room mic in Jérémy’s recording and his spot mics had been placed close to the instruments in a bid to achieve separation. This meant that the raw sound was dry, with the edgy quality you often get when affordable capacitor mics are positioned right up on a source; what natural reverb there was consisted of bright early reflections from the tiled floor.

Jérémy’s reference for the guitar sound was a ‘live in the studio’ performance by the group Selmer #607.Jérémy’s reference for the guitar sound was a ‘live in the studio’ performance by the group Selmer #607.

By contrast, Jérémy’s reference for the guitar sound was a ‘live in the studio’ performance by the group Selmer #607. The YouTube video doesn’t reveal whether room mics were used, but a Coles 4038 ribbon mic is visible on the lead guitar, there are more soft furnishings in evidence, and possibly more distance between mics and instruments. As well as a more mellow and rounded tone, Jérémy’s references thus had a more open, spacious feel than his own raw tracks. He’d tried to add this back in at the mix using artificial reverb, but it hadn’t quite worked.

Close Encounters

Taken on their own terms, though, there was a lot to like about the up‑front, dry quality of the raw tracks. Jérémy was clearly after a roomier sound, and I was keen to explore how this could be achieved, but it struck me as optional, and I wanted to deal first with any timbral issues that afflicted the basic recordings.

It makes a nice change to work on a mix with only four tracks! The guitar solos have been duplicated on a second track immediately below, and the stereo track at the top is Jérémy’s mix for comparison purposes.It makes a nice change to work on a mix with only four tracks! The guitar solos have been duplicated on a second track immediately below, and the stereo track at the top is Jérémy’s mix for comparison purposes.

The close mic placement had delivered good levels of separation, but there was a price to pay for this. The guitar sound was a little fierce, with lots of energy in the upper midrange and an emphasis on transients, pick scrapes and the like. Jérémy’s accordion was clearly well maintained and tuned, but the treble‑side mic had picked up a distracting ‘ticking’ noise from the keys; and though the tone from the treble and bass mics was decent enough in isolation, it changed significantly when they were combined.

There are some situations where we’re habituated to think about phase issues: when multimiking a drum kit, for example, or combining a mic and DI on the same source. But they also arise when you’re miking both ends of an instrument such as an accordion or concertina. What you have might look like a spaced stereo pair, but in practice, both mics are acting as spot mics on their respective ends of the instrument, and both are picking up spill from the other end. You’re very unlikely to want to pan these mics fully left and right at the mix, so it’s worth experimenting with polarity and time adjustment to get them to combine sweetly. In this case, I ended up reversing the polarity of the bass‑side mic and advancing it by about 1.5ms (which, not coincidentally, is about the time it takes sound to travel from one end of the instrument to the other).

I adjusted the relative polarity and timing of the two accordion mics to try to make them work as well as possible together in mono.I adjusted the relative polarity and timing of the two accordion mics to try to make them work as well as possible together in mono.

This produced a raw accordion tone that sounded pretty good to my ears, although it was a bit more strident than the accordion in Jérémy’s reference track. The tone also varied through the recording, as it will when you close‑mic an instrument that doesn’t stay still. Several bands of dynamic EQ spread across the midrange seemed to help somewhat, and I ran the treble‑side track through iZotope’s RX De‑click to tame the ticking from the keys.

Whereas the treble‑side mic was clean and the playing on that side more or less continuous, the bass‑side mic had picked up more spill from the guitars, and there were relatively long stretches where the left hand was idle. Since the spill didn’t sound all that great, I used the basic Pro Tools Expander/Gate plug‑in to drop the level in these rests, with a side‑chain filter set to focus on the low midrange. The accordion is not an instrument that produces endless overtones, and I decided that everything above 6kHz or so on the bass side was spill, so I cut it out with a low‑pass filter. Finally, having bussed both accordion tracks to a stereo aux channel, I experimented with some gentle dynamic control, eventually finding Sound Radix’s Drum Leveler more ‘invisible’ than an orthodox compressor.

Taking The Lead

The first half of the Charleston consisted mainly of an accordion solo, with Mathieu and Solal playing impressively tight rhythm guitar. In the second half, the accordion switched to rhythmic comping while Solal took over lead duties, before both combined to restate the main theme at the end. As long as the guitars were playing chords, spill was not really an issue, and my main concern was to soften the spiky sound a little. I used separate instances of Oeksound’s excellent Soothe plug‑in to tame the abrasive upper midrange on both instrument, and on Solal’s guitar I added a de‑esser targeting the 5‑8kHz range to take the edge off the transients.

The lead guitar was more of a challenge. No matter how hard the player digs in, it’s always a struggle to bring single‑note lines on a steel‑strung acoustic to the front of the mix. Spill from the other instruments was much more obvious on this part, and the combination of hard picking and close miking made the solo guitar particularly spiky. Up to a point, this isn’t a bad thing, but when gypsy jazz guitarists I’ve recorded have talked about the importance of ‘manouche bite’, they usually seem to be referring to the 2‑3 kHz area rather than 5kHz and above.

The guitar solo was the element that required the most extensive processing. Oeksound’s Spiff helped to tame the aggressive note attacks — it may look as though this setting is applying a high shelving EQ boost, but in fact it’s telling Spiff to apply more processing in that region.The guitar solo was the element that required the most extensive processing. Oeksound’s Spiff helped to tame the aggressive note attacks — it may look as though this setting is applying a high shelving EQ boost, but in fact it’s telling Spiff to apply more processing in that region.

I decided fairly early on that I would need to treat the solo section differently from Solal’s rhythm part, so I cut it out and duplicated it on a second track. It took quite a bit of experimentation to achieve the right balance of tonal warming‑up, transient softening, dynamic levelling and spill reduction. Oeksound’s rather unsung Spiff transient‑shaper plug‑in did a lot of the heavy lifting, with another instance of Drum Leveler tackling the dynamic control and spill reduction, and some dynamic EQ toning things down from 5kHz upwards. During the accordion solo sections, it seemed natural to have the rhythm guitars hard-panned, but I positioned this lead part more centrally: not so much as to create a jarring jump at the transition, but enough to make clear that it was now the boss.

Sound Radix’s Drum Leveler is a very effective plug‑in for controlling the dynamics of any percussive source, including solo lines on acoustic guitar.Sound Radix’s Drum Leveler is a very effective plug‑in for controlling the dynamics of any percussive source, including solo lines on acoustic guitar.

To The Limit

By this stage, I had dealt with most of the issues that had been bothering me about the individual sources, but I hadn’t made any radical changes to their sound, and I certainly hadn’t come close to matching Jérémy’s references. The aforementioned spaciousness was still absent, but also, the mix still sounded rather thin in comparison. It was hard to know how much could or should be done about this — obviously, a track with no bass instrument is never going to sound as full as one with a double bass — but at the very least it was clear that any master bus processing should be reinforcing the low midrange rather than accentuating the top end. This was another area where I auditioned lots of different plug‑ins, eventually using two tape emulations in series to add a little thickness.

Transient‑rich material like this is also an interesting test for limiter plug‑ins. A fair amount of limiting was needed to match the level of Jérémy’s mix, and was quite beneficial in terms of smoothing out the spikiness, but I was surprised at how different the plug‑ins I tried sounded from one another. In the end I felt that Lo Punch and Dynamic Perception from Slate’s FG‑X seemed to add something that wasn’t available from other processors.

Room For Improvement

And so, finally, to the all‑important reverb. Or was it? Jérémy’s references did indeed have more ambience than his own recording, but it was dark, moody and very controlled: the very opposite of ‘reverb as an effect’. I don’t know how much of it came from the source mics and how much was added later at the mix, but it certainly didn’t sound ‘pasted on’. Nor did it sound much like the reverb Jérémy had chosen for his own mix.

To counteract the inherent brightness of the recording, my main reverb was a very dark ambience patch, which I modified to make it even darker.To counteract the inherent brightness of the recording, my main reverb was a very dark ambience patch, which I modified to make it even darker.

I spent a lot of time trying out different reverb treatments to get close to the same feel. Fairly quickly, I decided that the main ‘bread and butter’ reverb should be a short, dark ambience patch that would thicken up the low midrange without sounding boomy or drawing attention to itself in the treble. This helped to add weight to the rhythm guitars and make things gel, but it still left the overall mix sounding quite dry in comparison with the references.

If there was a way to use artificial reverb to recreate the more ambient feel of the references, I never found it. But in the end, we both came to feel that this actually wasn’t a problem.

One of the blessings, or possibly curses, of having only close mics to work with was that this made it possible to try additional reverbs tailored for the solo instruments. My initial effort was quite conservative, and much dryer than Jérémy’s own mix. Initially, his response was that he felt it was too dry, but after several further attempts, we agreed that it had probably been the right approach, or at least the best we could hope for. If there was a way to use artificial reverb to recreate the more ambient feel of the references without just sounding like someone had added a ton of reverb to otherwise dry tracks, I never found it. But in the end, we both came to feel that this actually wasn’t a problem. With the possible exception of the guitar solo, any tonal issues with the source mics were relatively minor, and once these had been addressed, they came together to deliver a punchy, up‑front mix that worked well on its own terms.

Reference Class

I’ve name‑checked some expensive third‑party plug‑ins in this article, but that’s really just because those are the plug‑ins I’m familiar with. This was fundamentally a decent recording, and it would have been perfectly possible to do a good mix with the bundled plug‑ins in any DAW. In fact, if you listen to the isolated parts I’ve provided as audio examples, you’ll hear that what I did to the raw parts was mostly quite subtle. Simply pushing up the faders on the raw tracks would actually have produced a respectable mix. The problem lay, above all, in trying to make it into something it wasn’t.

If there’s a lesson to be learned from this Mix Rescue, it’s that although references are important, how you use them is even more important. Anyone can identify a slick commercial track and think how nice it would be if their own mix sounded like that; but if you get too hung up on trying to do so, you can end up with something that is neither fish nor fowl.

The sound of a simple live recording is baked in at tracking, and there’s only so much that can be done to shape it at the mix.

In this case, the horse was already trotting across the countryside and the stable door swinging open by the time Jérémy had started to think about references. The sound of a simple live recording is baked in at tracking, and there’s only so much that can be done to shape it at the mix. If there was a time to have a reference sound in mind, it would be when positioning the mics and the players, though of course it’s not always practical to hold up a session to make comparisons. For the next session, I’d suggest trying a little more distance between players and mics, and getting hold of a ribbon mic for the lead guitar. If time and resources permit, it might also be worth experimenting with a more traditional approach where spot mics on the instruments are used to augment a main stereo pair that captures the band as an ensemble.

As Jérémy and I batted mixes back and forth, we both came to the conclusion that trying to match his references was not only impossible but counter‑productive. Not only do you risk losing what’s good about what you have in pursuit of something you’ll never attain, but you also undermine your own confidence as an engineer. References can be an invaluable tool as long as they remain just that: a tool. If you let them become an end in themselves, rather than a means to an end, disappointment is inevitable!

Check out the audio examples at sosm.ag/mix-rescue-0521 to hear the original and rescued mixes. Or download the high-res 24-bit Wav ZIP file and audition them in your own DAW.

Package icon mixrescuegypsyjazz.zip

Rescued This Month

Accordion player Jérémy Dutheil and guitarist Solal Poux are French gypsy jazz specialists, who gig regularly in duo, trio and quartet formation when such things are allowed!

Remix Reaction

Mix Rescue: Jérémy DutheilJérémy Dutheil writes: “I was a little lost with my mixing process, trying to sound like my references but unable to put words on what I liked about their recordings. I knew we had a pretty solid song in terms of playing, but couldn’t make it sound right... It was very frustrating!

"Sam succeeded in clarifying things with his mix, bringing each instrument to its own place. I eventually realised that I was focusing too much on some features, especially reverbs, that ultimately weren’t necessary. Sam exchanged a lot of emails with me during the process, which was very interesting! Thanks a lot — I’m ready to try to reproduce his work on other songs!”