Oeksound Bloom

Oeksound’s long‑awaited third plug‑in is a quietly radical alternative to equalisation.

Some plug‑in developers seem to release a new product every week. Others are more selective — and they don’t come much more selective than Oeksound. Launched in 2018, the enigmatic Finnish coding wizards’ first plug‑in quickly became one of those tools that every big‑name engineer seems to have in their kit, and for good reason. In an era when we often work with phone demos and ropey home‑recorded tracks alongside pristine studio material, Soothe’s unique ability to dial back unpleasant resonances and alleviate harshness makes it invaluable.

The success of Soothe perhaps meant that Oeksound’s follow‑up, Spiff, slipped under some people’s radars, but as I’ve got more of a handle on how it works, it too has become one of my favourite dynamics processors. It offers a novel, frequency‑dependent approach to transient control which is completely different from other ‘transient shapers’. Since then, there’s been a version 2 of Soothe, and a low‑latency derivative optimised for live sound, but it’s been more than five years since Oeksound launched an entirely new plug‑in.

At this year’s NAMM Show, however, Oeksound finally unveiled their third major product. And, spoiler alert, it was worth the wait.

What It’s Not

Oeksound describe Bloom as doing “What we wish an EQ would do,” and this, it turns out, isn’t just one thing. Like an EQ, it has corrective uses, but also creative applications, allowing the timbre of recorded audio to be modified. However, Bloom is not an EQ. Nor, despite similarities in the user interface, is it a multiband compressor. And although it has something in common functionally with some products based on machine learning, such as Sonible’s SmartEQ, it’s a purely algorithmic design. So what is an “adaptive tone shaper” when it’s at home?

There’s a certain level of mystery surrounding exactly what Bloom does, but a few things are clear. Firstly, unlike Soothe, it’s essentially a broad‑brush process: rather than notching out hundreds of tiny resonances, the tonal changes it applies typically span an octave or more. Second, it’s a dynamic process, in the sense that it is constantly adjusting what it does in response to the source. Third, it can also be dynamic in the sense of changing signal dynamics, but unlike a compressor, its effect is for the most part level‑independent.

Bloom is a native plug‑in available for macOS and Windows in all the usual formats. It’s authorised using the iLok system, but a physical dongle is not required. It occupies a relatively compact window tinted an attractive shade of rose pink, with no extra tabs or panes, and there are just five main controls. The large Amount control is self‑explanatory, but the four Tone Control sliders that adjust specific frequency ranges are less so. For one thing, they’re actually X/Y pads rather than conventional faders: moving the central handle up introduces a ‘boost’ in that signal range, and pulling it down initiates a ‘cut’, but you can also drag the handles sideways to set the centre frequency of the band. This makes it seem superficially like a multiband compressor, but in practice, it’s very different.

I’ve placed the terms ‘boost’ and ‘cut’ in inverted commas because it’s important to understand that they don’t apply gain changes in any normal sense. It would be more apt to think of them as offsets that can be applied to the core process: the settings of the sliders adjust the ‘target’ response that Bloom tries to nudge your sound towards. Crucially, leaving all the sliders at zero doesn’t mean that no processing takes place: it means that the processing will aim to push your sound towards a balanced target. By contrast, pushing the high and low sliders up and the middle ones down will define a ‘scooped’ target, and so on.

Line Dancing

Bloom’s processing is represented in an animated ‘processing graph’ which occupies the space beneath the Tone Control sliders and has amplitude on the vertical axis and frequency on the horizontal. Turn the Amount dial right down to zero, and the graph looks like a flat line. As you turn the Amount up, the line begins to deform and move around in response to audio input. Where Bloom’s algorithm decrees that a boost should be applied in a certain area, this shows up shaded white, while cuts are indicated by the line falling below the horizontal and eating into the solid mauve region. Additional cuts or boosts prompted by the slider settings are shown in the colour of the relevant slider. Draggable handles at either end of the processing graph allow you to introduce low‑ and high‑pass filters.

The Amount control ranges from zero to 10. As you push it up, Bloom becomes increasingly assertive in the boosts and cuts that it applies, but the overall subjective level remains constant. A Wet Trim setting lets you adjust the level if for some reason this doesn’t happen. This is also useful if you want to employ Bloom as a parallel processor, and when you push the Amount control beyond 7. At this point, the white boosted areas on the processing graph start to hit the ceiling, introducing a cool and easily audible compression effect. Again, this isn’t the same as conventional broadband or multiband compression, but it imparts a similar sort of pumping, breathing quality to the signal. This part of the processing is level‑dependent, and a Squash Cal control has a function somewhat similar to that of the threshold setting on a normal compressor.

The Attack and Release controls govern the speed with which Bloom reacts to changes in the audio. Like most Bloom parameters, they are arbitrarily calibrated from zero to 10, and it would probably be an oversimplification to call them ‘time constants’. Their settings are often more obvious from the movement of the processing graph than from the actual sound; at least until you reach the squash range, it remains smooth at all settings. Bloom can operate in standard or high‑quality modes, the latter incurring a greater CPU overhead but making an audible difference on some sources, especially complex material. It can also be switched into a low‑latency mode for tracking, which reduces the look‑ahead delay to 1.33ms at 48kHz at the standard quality setting.

Finally, if you instantiate Bloom on a stereo track, it defaults to applying the same processing to the left and right channels, presumably responding to the sum or average across both. However, it’s possible to switch it into Mid‑Sides or dual‑mono modes, and if you do this, each Tone Control band sprouts a second slider, allowing you to set different targets for left and right, or for the sum and difference channels.

On stereo tracks, Bloom gives you the option to process left and right or Middle and Sides signals independently.

Rose Tinted

If describing what Bloom does isn’t easy, then neither is describing how it sounds. In the most general terms, the best I can come up with is to say that it acts like a magic mirror, reflecting back a version of the source sound that is in some indefinable way more attractive. Or perhaps it’s the audio equivalent of the filters that make us look younger and more beautiful in social media videos. If you’re coming at it with a mindset formed by EQ and multiband compression, it takes a little while to get used to the fact that Bloom can have a powerful effect even with all the Tone Control sliders at zero. Simply insert it on a track and you will hear an immediate change in the sound — one which is likely to make you think “Wow”. Within two minutes you’ll have forgotten it’s there, and it’s only when you bypass it once more that the harsh reality of your recording comes crashing to your attention.

In theory, if you feed Bloom an unchanging signal that’s already well balanced by its own lights, any movement on the processing graph should be minimal, or at least equally spread across the spectrum. To get a handle on what it does, I tried using pink noise as a source, since some people recommend this as a balanced target spectrum for mixing. Doing so made clear that Bloom’s idea of a balanced tonality is more ‘scooped’ than the spectral balance of pink noise, because it tried to push up the bass and high frequencies and make a broad cut in the midrange. At the same time, thankfully, its target is clearly much less bright than white noise.

However the target is defined, it works, because I don’t think I have ever encountered another plug‑in where the default preset was so universally applicable. As soon as you instantiate Bloom, you’ll hear your audio being subtly massaged towards Oeksound’s target frequency balance, and it’s rarely the worse for it. Whatever this target balance is, it works equally well on anything from individual sources to the master bus, and from delicate acoustic recordings to brutal electronic kick drums.

As I’ve already mentioned, the process is incredibly smooth, and unless you stray into the ‘squash’ range, you’re unlikely to notice any changes to the dynamic behaviour of your signal. On a drum loop, for example, the audible effect of the Attack and Release controls has more to do with changing tone colour than with bringing up or down the sustain portion relative to the transients.

...it’s a remarkably effective alternative to equalisation for tonally variable sources like the human voice, and especially voices afflicted by bad room sound, poor mic technique or inappropriate mic choice.

Flower Festival

The applications for this plug‑in are endless, but I’ll list a few highlights that I encountered in my testing. First of all, it’s spectacularly good at dealing with damaged or badly recorded audio. In restoration work, it very often achieves in seconds what you could spend hours trying and failing to do with EQ or multiband compression. Bloom is at its most impressive in those apparently impossible situations where you need to reshape the tone of a track or source but doing so with EQ seems to bring out all sorts of nasties that the wonky timbre previously obscured.

Second, it’s a remarkably effective alternative to equalisation for tonally variable sources like the human voice, and especially voices afflicted by bad room sound, poor mic technique or inappropriate mic choice. The default target profile generally drags raw voice recordings towards a brighter sound, filling out the upper midrange whilst controlling wooliness and proximity boost in the low mids. Yet, when Bloom encounters sibilants and other consonant sounds that already have plenty of high‑frequency energy, its adaptive nature means these are left well alone or even reduced. The upshot is that you get a more tonally consistent vocal sound, with no need for de‑essing and no risk of introducing side‑effects such as lisping.

Third, it often enables you to apply a greater degree of tone shaping than is possible with other tools, whilst remaining natural. For example, I was able to take a drum overhead track recorded with a vintage ribbon mic, push up the High and High Mid Tone Control sliders and produce something that sounded for all the world like it had been tracked with a capacitor mic. Attempting a similar transformation with EQ just made everything sound, well, equalised. Or rather, badly equalised.

Fourth, Bloom naturally saves you from yourself in a way that an equaliser won’t. I’m sure we’ve all had the experience whereby we boost something a little bit with EQ, like the results, crank the boost a little higher, and end up losing our mental reference as to what that source or mix should sound like. The consequence is often an instrument or mix that sounds harsh, brittle, thin, boomy or otherwise unbalanced, and doesn’t translate well between systems. Because Bloom is always pushing things towards a reference spectrum that is intrinsically balanced, it’s much harder to fall into this trap.

Wish Fulfilment

In short, then, Bloom is probably the closest thing I’ve come across to the proverbial ‘make it sound better’ plug‑in. Are there no down sides? Well, there are certainly sources where static EQ is preferable to my ears: on heavily distorted guitars, for example, Bloom seemed to lose some of the solidity and substance of the sound. And there will always be times when you need to use EQ to deliberately make something sound pokey, aggressive or otherwise unbalanced; using the Tone Control sliders can force Bloom into adopting an unbalanced tonal target, but it won’t deliver the bite you get from pushing the midrange on an API. The ‘squash’ effect is interesting and characterful, but I didn’t find a real‑world use for it during the review period. And there were times when I wanted to be able to process only part of the frequency spectrum, which isn’t currently possible.

Bloom is probably the closest thing I’ve come across to the proverbial ‘make it sound better’ plug‑in.

Most of all, though, because Bloom’s processing is so seductive, it’s alarmingly easy to apply more than you need to. Like Soothe, it almost always makes whatever you’re applying it to sound nicer in and of itself; but that doesn’t always mean it works better in the mix. A timbrally balanced mix isn’t normally achieved by making every individual element timbrally balanced, but by playing the contrasting tonal imbalances of different sources off against each other, and there are times when sounds need to be unbalanced and harsh in order to fulfil their role in the bigger picture. Hence, as I experimented with Bloom, I became aware of a slightly paradoxical aspect to its operation. On the one hand, it is uniquely transparent, in that it can make timbral changes sound natural that would be impossible with EQ or other tools. On the other, that means it can actually have a sound of its own, in that if you Bloomify too many of the sources in your mix, it will begin to acquire a sort of homogenised, ‘too good to be true’ quality. This outcome is rarely something you’re consciously aware of, more an uncanny feeling that lurks in the back of the mind, and as such, is easy to overlook.

But at the end of the day, what this boils down to is really just that it’s possible to over‑use this plug‑in — and if that’s a criticism, it’s one that applies to every plug‑in ever made. Self‑restraint is needed to apply any kind of audio processing, and if Bloom requires more self‑restraint than most, it’s because its effect is so appealing.

Equalisation is one of the most fundamental tools available to the audio engineer, and it’s something we take for granted in tracking, mixing, live sound, restoration, mastering and every other context. Oeksound have set out to make Bloom the processor they wish EQ was — and, consequently, it has an equally broad range of possible applications. This is a unique plug‑in that has something to offer everyone from broadcast engineers to film dubbing mixers to mastering engineers, from the most humble of home studios to the most advanced mixing suites. Bloom really is pretty special, and if you don’t believe me, sign up for the 20‑day free trial!