Ableton Live 12

Live 12 introduces a raft of new features, devices and general improvements.

Hot — or at least warm — on the heels of the latest version of the Push controller (Sound On Sound, July 2023) comes a new version of Ableton’s flagship DAW, right on time for the roughly three‑year refresh cycle. The much‑trailed new feature is support for alternate tuning systems, but there’s also a bunch of improvements to Live’s interface, some nifty new editing features and a handful of brand‑new devices. Let's explore...

Interface Improvements

Live has always had a pretty regimented idea of what should appear where in its window (or both windows, if used in two‑window mode). That regime has been broken down a bit in Live 12. For a start, the mixer view from the Live Session can now be shown in the Arrangement, a design move that makes such obvious sense one wonders why it wasn’t done before. And — sacrilege! — you can view a track’s device chain and selected clip at the same time. Try to do both of these things at once and you’re probably going to run out of editing space on a laptop screen, unless you scale the view down and have very good eyesight, but on even a modest external screen these are clear usability improvements.

Navigating around a Live Set has been made easier too. There’s a new Navigate menu that exposes keyboard shortcuts for selecting the main areas of the interface, and various Tab key combinations that then walk around controls within each area: along the control bar, up and down the controls in a mixer channel, and so on — or even within a single device. I’m a fan of keyboard navigation, but there’s so much on screen in a Live project that I’d probably wear my Tab key out if I tried to use it exclusively to navigate. You will probably end up learning a handful of shortcuts and then switching back and forth between keyboard and mouse while working.

Browser

Filtering by content and device type.Live’s browser has had a big revamp. A major new feature is filters: each filter is a set of on/off tags for specific characteristics of many types of object within Live, from an entire Live Set all the way down to a single clip file. The actual filters available change depending on the type of object: for samples the tags include content (sample or video), type (loop or one shot), category and character of sound, and so on. For plug‑ins the content tag can be device or preset (although plug‑ins rarely export presets to Live unless you save them to disk), and there are tags for properties such as format (AU, VST2, VST3) and creator (the vendor).

At every stage the filter is being applied to the root selection (in Library or Places) in the left column of the browser, not the actual item selected on the right. And in any specific library location or place, the tags are gently highlighted according to whether there are any items at all that match them, according to the current active filter, guiding you to finer searches with more tags.

Filters can be edited, so that you can add your own tags — the editing pane is also where you get to change which tags are enabled on the selected object. I added my own name to some Max For Live devices I’ve created, and was pleasantly surprised to find that I could move a device from one folder to another without its tag disappearing. The tagging appears to be done by name: change the name and the tags vanish.

I don’t know whether I will spend an afternoon at some point going through all my projects adding tags — I suspect not — but I found it immediately useful to filter plug‑ins by format, to make sure I was consistently using the VST3 versions rather than VST2 or AU, since Live 12 no longer groups the different formats into folders. Tagging entire Set files might well be useful for some workflows.

For a more contemporary take on searching, presets and samples support ‘Sound Similarity’ searching, using tags and neural net analysis to list content deemed similar. Results varied from spookily similar to vague approximation, so at the very least another way to influence your creative choices.

Live still features ‘collections’, which are coloured labels attached to objects. These seem to be orthogonal to the new tagging machinery. Select a collection by colour, and its contents can still be filtered. In any view, filtered or not, collection colours are shown against the items.

One thing I found myself wanting to do was search the currently loaded Live Set or Project for devices it was using — useful for swapping out those obsolete VST2 or Intel‑native plug‑ins — but this seems not to be possible for anything embedded in a Set, except for Max For Live devices, which exist separately on disk. Other DAWs do allow searching and listing of currently loaded devices, and I miss that feature in Live.

Tags are global, and appear to be stored in the application’s preferences folder, so if your tags are important, be sure to back them up.

Modulation

One of the prime features of Max For Live, apart from generating and processing MIDI and audio, is its ability to control the Live Set that it runs inside. This enables you to build and use modulation sources like LFOs and envelope followers, and attach them to controls in other devices in the Set, or even mixer controls. Until now, this feature came with a down side, in that it wasn’t doing modulation in the true sense at all, but something akin to automation take‑over. A parameter under the influence of Max For Live could be changed in value, but the value would be changed permanently. This would potentially mess up any preset being saved, and also preclude any kind of manual update or automation of the same parameter. (You might, for example, want to have a long slow automation lane to open a filter, while also applying some LFO modulation.)

Two Max For Live LFOs modulating a mix.

‘Real’ modulation already exists in Live: clips can contain envelopes which either modulate or automate parameters in a track’s devices or mixer. But now, Max For Live can do proper modulation as well, offsetting a parameter from a base value. This modulation can be unipolar or bipolar, and its presence is indicated by a green dot next to the parameter being affected. It is possible, as the illustration shows, for a parameter to be under both automation (red dot) and modulation (green dot) control at the same time.

I discovered that it is also possible to modulate a parameter with a clip envelope and a Max For Live device at the same time, and the effect appears to be a combination of the two, but a parameter cannot be controlled by more than one Max For Live device. (I’m sure some kind of clever modulation‑mixing device could be built if there were some demand for it.)

Plug‑ins can have modulation applied to them as well, but this is just treated like automation, and changes the actual value of any parameter being controlled, no doubt a restriction of the current plug‑in standards (VST and AU). If Ableton ever support the newer CLAP plug‑in format, the problem might well go away for some plug‑ins at least.

Scales & Tuning

Live is now scale‑aware, in that clips and MIDI effects, as well as some devices, can have their notes constrained to a particular scale rooted in a particular key. Click a toggle in the control bar and that scale will be the default for clips and devices (and the new MIDI Tools, described below). Clips will show the active scale in the piano roll, and you can hide MIDI pitch rows which aren’t in that scale. MIDI devices can follow the current scale as well: the Scale device is a natural contender here, able to constrain notes to the current scale rather than one set in the device itself; the Arpeggiator can do this also.

The current scale highlighted in a clip.

At this stage I started to wonder how I would go about writing a track which modulated key part way through, a technique not unknown even in modern popular music. Even though the main scale selector is shown prominently in the control bar, this setting tracks the scale property of the selected MIDI clip. When a clip is launched in the Session, or starts playing in the Arrangement, its particular scale (if any) takes effect and is imposed dynamically on any devices in the track which are set up to use scale selection.

The global scale setting in the control bar.Instruments can potentially make use of scale information as well, although to date Meld (see below) is the only one to do so. When it’s in scale mode, Meld’s two oscillators can be tuned in scale degrees rather than semitones, so that (for example) four scale degrees in C major delivers a flat fifth on the B but a perfect fifth elsewhere, as per the scale. Meld’s filters are also scale‑aware: the plate and membrane resonators will track scale pitch, at least to an extent.

Older instruments, such as Operator, Drift and Analog, have no awareness of scales themselves, although they will play scale‑aware clips. Max For Live also has no awareness of the scale system, at least for now — I’d expect that to change in short order.

Live’s support for scales results in MIDI notes being filtered or shifted by semitones as they are processed, but has no impact on the actual frequencies associated with the notes: by default, everything is in equal temperament with an A above Middle C of 440Hz. You are welcome to detune individual instruments, and many third‑party plug‑ins and hardware instruments support alternate tuning systems, but it’s rare in DAWs — a quick Google search suggests that Apple Logic Pro is the only other major platform to offer global tuning to date, although Bitwig Studio has a device for generating microtonal note pitches.

Scales activated for two MIDI devices.

This omission seems odd when you consider that the standard equal temperament tuning system is a 400‑year‑old compromise which allows a single tuning to support music written in any key while guaranteeing the same frequency intervals between notes, the cost being that the harmonic intervals are slightly out of tune. Retuning a physical instrument between songs is not very practical (for keyboardists anyway; guitarists do it), but retuning a software instrument, or a digital hardware one, should be straightforward and immediate. In the West we are used to equal temperament and perhaps don’t fully realise what we’re missing, but a listen to Bach’s Well Tempered Clavier, in which a piece of music in every possible key is played in a single tuning where note intervals differ between keys, serves to refresh the ears. For more modern and radical tunings, check out the work of Wendy Carlos, Arnold Schoenberg and Conlon Nancarrow.

This discussion skirts the issue that standard 12‑note‑per‑octave tunings dominate in Western music to the detriment of other tunings in the world. Tools which support other tuning systems are supporting musicians whose creative work is potentially compromised by the traditional equal scale — and opens a door for us to experience and experiment with such musical ideas too.

Live 12 supports arbitrary tuning systems (in other words, mappings between incoming MIDI note numbers and frequencies). All of Ableton’s instruments are able to adopt a Live Set’s tuning, and Live seems able to cajole MPE‑capable third‑party devices to retune themselves as well. (ROLI Equator 2 followed Live’s tunings accurately, at least to my ears.) For non‑MPE plug‑ins, you’re probably out of luck, unless the device can be retuned independently.

Some tunings in the browser.Live’s browser allows a single tuning at a time to be activated in a Live Set, although it can be customised or disabled per track. The tunings are tagged and can be searched by filtering. There is a selection of Arabic maqam scales, some just intonations (12 notes per octave and above), a selection of EDO (equal division of the octave) tunings, and Wendy Carlos’ three experimental tunings (alpha, beta and gamma), which are notable for not having repeated octaves. You can also add your own tunings as Scala files — there is at least one web‑based environment for designing and exporting files, which Live 12 seems to read without any problems.

If you work with Max For Live, direct tuning system support is absent (although I would expect this to be remedied before too long), but if your device is MPE‑capable then Live seems to treat it like any other MPE device and sends per‑note pitch variation to produce the correct pitches in the current tuning. A quick bit of Max For Live programming seemed to confirm this, and I tested the MPE‑capable Granulator III (see below) with various alternate tunings: it seemed to match the pitches correctly.

New Devices

Live 12 Suite sports two major new native devices — one instrument, one effect — plus the return of a Max For Live favourite. The native devices fit into the 170 pixels of height afforded by the device view pane, but also have an expanded view which houses each device’s modulation matrix. The idea of a mod matrix as first class citizen is relatively recent: of Live’s instruments, only Wavetable (which arrived in Live 10) has anything similar, and the matrix there doesn’t feature in the expanded view. I’m personally all in favour of modulation getting a more prominent billing.

But we’re getting ahead of ourselves. The new instrument is called Meld, and Ableton’s release documentation doesn’t really categorise it. In fact, its architecture is relatively conventional: two voice paths layered in parallel, where each comprises an oscillator and filter, with envelopes, LFOs and a tone control. So far so straightforward, although the design does, as Sherlock Holmes might say, present some points of interest. There is a choice of 17 different filter algorithms, including phaser, comb, vowel and two resonators. The first of the two LFOs per voice has a dedicated ‘effects’ section for shaping, forming a little modulation system of its own before the main modulation matrix comes into play. And most notably, the oscillators are, in a sense, semi‑preset: you pick one of the 24 oscillator types from a menu, and then have a grand total of two macro knobs to customise what that oscillator does (over and above the usual octave, semitone and cent tuning controls). The names assigned to the macros change depending on oscillator type (although they’re currently still referred to as ‘Osc Macro 1’ and ‘Osc Macro 2’ in the modulation matrix).

Meld, with its oscillator types and modulation matrix.

I’ve summarised the different oscillator types in the ‘Meld Oscillator Types’ box. They include a selection of basic waveforms, granular sources over different waveforms, noise, square wave with hard sync, a few varieties of two‑operator FM, a warbly chip‑tune source, a climbing (or falling) Shepard tone, and more. They are all smooth and rich‑sounding, and surprisingly versatile given the minimal number of controls. And for extra fatness (as we used to say in the 1990s) there’s a unison mode, with voice spread available as a modulation source.

At first I thought the ‘preset’ nature of the oscillators would be a serious restriction, but each type is well‑crafted in terms of its macro control, and the multi‑mode filters downstream are more than flexible enough to provide powerful sound shaping. The LFO modulation system is sophisticated, and, as a bonus, the envelopes can be looped, so it only takes a bit of messing around with the modulation matrix to get into moody VCS3 territory, if that’s your thing. Dial up FM for both oscillators and you’re effectively doing a subset of four‑operator FM. Use the voice paths separately for attack and sustain parts of your sound and you’re into the once‑fashionable ‘linear arithmetic’ layering of the Roland D‑50. Overall, I like Meld a lot — it looks simple, but its well‑crafted component parts quickly assemble themselves into surprisingly versatile generators of sonic material.

Roar

The Roar effects device is described as providing colouring and saturation. It consists of three waveshapers, each of which is equipped with a dedicated filter which can be positioned pre‑ or post‑shaper. The shapers can be arranged in a number of different signal paths: single, serial, parallel, multiband, Mid‑Sides, feedback — a handy little layout icon and some dedicated level meters show what’s going on. Oddly, multiband is the only mode where all three shapers are available — in the other modes, it’s just the first two (and in the case of ‘single’, just the first one). The waveshaper configurations range from the tried and tested (tubes, preamps, rectifiers) to the gritty (‘digital clip’, ‘bit crush’) to the outlandish (‘fractal’, ‘tri fold’). The filters are similarly versatile, from the usual low‑, medium‑ and high‑pass to a nice multi‑mode morphing configuration, a comb filter and one for aliasing‑packed sample‑rate reduction. More shaping is provided by drive and tone controls upstream of the waveshapers, while a shared audio section with band‑pass filter and delay is routed in to provide feedback options. A compressor at the end of the audio chain allows the final signal to be reined in if required.

Roar, in all its graphical glory.

I have to admit that I’m not a massive fan of digital waveshaping, and have never bit‑crushed anything in my life, much preferring subtle and soft saturation effects. Many of the Roar presets are glitchy, digital and harsh to my ears. However, with a bit of experimentation it’s possible to uncover some effects that are a bit more gentle and melodically engaging. The comb filter is useful for bringing in some tonal heft, as is the feedback delay, which can be configured in units of MIDI note pitch. My personal tastes aside, this is a device which guitarists will probably love. And, as with the Meld instrument, the built‑in modulation matrix is a significant asset. It sports two LFOs, a noise modulation source (with sample & hold, Brownian motion and others), and an envelope follower. Apply Roar’s processing through modulation rather than just turning up the knobs, and it’s possible to get some subtle and engaging effects, especially with rhythmic material.

Granulator III

Live 12 Suite comes with the third generation of Robert Henke’s granular sample player, Granulator III. The premise is largely unchanged from previous versions, but with quite a few refinements. I fired up Granulator II for comparison, and noticed some things that had been lost: version 2 had an FM processing section, dual filters and EQ, where version 3 loses the FM and has a single filter, albeit with more filter modes. Henke tells me that having a single, more powerful filter makes modulation easier (the modulation system is much enhanced over the previous version), while FM was not heavily used.

Robert Henke’s Granulator III.

FM and filters aside, Granulator III is seriously enhanced compared to its predecessor. It also looks and feels convincingly like a native Live instrument, with a neater, much improved user interface. The top half of the display shows either a waveform with the granular ‘play heads’, or parameter editing controls, depending on which part of the voicing system is being edited. This view is selected as a side‑effect of clicking on a knob at the bottom of the panel: the sample or control view appears as appropriate (although either view can be locked), and a thin black strip of numerical values across the middle updates dynamically to allow modulation of the highlighted parameter. It’s not quite a modulation matrix, but is very neat and usable. (At some stage it would be great if Max For Live were to support full modulation matrix interfaces — perhaps we’ll see that in time for Granulator IV?)

Granulator III is MPE‑compatible, handy if you want to add per‑note expression or modulation to your granular voices, and it also means that it can participate in Live’s alternate tunings.

You can drag and drop audio files into Granulator’s waveform panel to replace the audio source, and the instrument is also capable of live sampling up to eight seconds of audio from another source such as a track or external input (or, more accurately, it can retain the previous few seconds of audio instantaneously). I discovered that you can even attach an LFO to the ‘capture’ button to periodically replace the buffer while it’s active, great for those experimental noise gigs. Captured audio can be saved as a file in the project to be reloaded later.

Analog & Friends

Meld, Roar and Granulator III require a Live Suite licence, but if your budget only stretches to the Live Standard edition, there are still some goodies which come with Live 12: Analog (virtual analogue synthesizer), Collision (simulated mallet percussion), Electric (electric piano), Tension (string instrument emulator) and Corpus (resonant object modelling effect) arrived with Ableton Live Suite 7 a good 15 years ago (SOS February 2008), but will now be part of Live Standard.

MIDI Tools

Live already has a toolbox of MIDI processing devices which, generally speaking, take in notes in real time and put out processed notes in response. These MIDI Effects are, in a sense, transient: they transmit their output but don’t save it, and any MIDI clips that feed them are unchanged whether MIDI Effects are running or not. The source clips and the final emerging MIDI are decoupled: the clip can be edited whilst effects are running, and a MIDI Effect can be reprogrammed — or even automated — while a clip is looping. The similarity to audio effects is obvious: the effect does not alter the original source or recording, and can be modified at any time. The humble Arpeggiator is a stereotypical MIDI Effect: it’s generally fed long chords and outputs a faster repeated ostinato of notes.

So far, so familiar. Live 12 brings some new MIDI processing to the table in the shape of MIDI Tools, but these work in a completely different way to MIDI Effects. The tools operate at the clip level (Session or Arrangement), destructively altering notes in the clip. There is no live processing of notes: MIDI Tools alter the pitch, placement and duration of notes in‑place in the clip, as part of the editing process (and there’s no automation control available).

The tools exist in two varieties: generators and transformers. The generators create notes directly inside a clip, whereas transformers alter notes that are there already. Both types of tool operate on a set of notes that are currently selected for editing in the clip: a transformer alters the selected notes, while a generator creates notes when first invoked and then alters them on subsequent invocations. By ‘invoke’ I mean a click on an ‘apply’ button that does a one‑shot application of the generator or transformer. The result is a new set of selected notes, but the tool can still be edited and reapplied if the effect is not quite as desired, resulting in a new, selected, note set.

An example might clarify things. The Arpeggiator (a transformer) will replace all selected notes with shorter, arpeggiated notes — and these will still be selected and active. Changing the arpeggiator settings will alter these notes, moving them, adding new ones or taking notes away, but keeping them all selected. Click elsewhere in the clip area to deselect the notes, and the spell is broken: you’ve effectively frozen the notes into the clip, as if you’d put them there manually (or recorded the output of a ‘real’ arpeggiator). Once notes are deselected, there’s no going back, except via global undo. Certainly, undo is your friend here when doing anything with the MIDI Tools — you’ll use it a lot — but do something like alter the settings in an instrument and you’ll clock up further undo steps, so you won’t be able to undo just the MIDI Tool actions.

There are clearly some subtleties to MIDI Tools, and the drawback that a tool isn’t ‘live’: once it’s been applied and the resulting notes deselected, you’re committed. On the other hand, you can then go on to edit the notes by hand, or for that matter apply other tools. And, creatively speaking, committing to a course of action is often a good thing.

At the time of writing, there’s a small set of bundled generators and transformers. In the generator category there’s Rhythm (generate grid‑quantised notes of a single pitch — best applied multiple times, one per drum part), Seed (a more general note‑maker with variable pitch, duration and velocity), Shape (turns a user‑drawn graph into pitches), Stacks (a chord sequence generator) and Euclidian (as the name suggests, a generator of four‑voice Euclidian step sequences).

The Euclidian MIDI Tool.

The transformers comprise the aforementioned Arpeggiator, Connect (connects equally pitched notes up in time, adding ties to taste), Ornament (adds flams and grace notes), Quantise (a note quantiser), Recombine (permutes notes in time), Span (alters note durations and spans notes to those following), Strum (guitar strumming), Time Warp (applies speed/tempo curves) and Velocity Shaper (applies an arbitrary graph shape to note velocities over time).

Euclidian and Velocity Shaper are actually implemented in Max For Live, rather than coded internally in Live like the others, and you can actually build your own MIDI Tools from scratch. The Max For Live devices are pretty much indistinguishable from the native ones, but you need to have an active Max For Live licence to access them, and they are not quite as aware of Live’s new scale mode as the others.

The overriding impression is one of refinement: interface improvements (particularly the browser), alternate tunings, powerful MIDI creation tools and a few tasty new devices.

Conclusions

Live 12 is a solid upgrade, and to me the overriding impression is one of refinement: interface improvements (particularly the browser), alternate tunings, powerful MIDI creation tools and a few tasty new devices — not to mention the arrival of a collection of previously Suite‑only devices into the Standard edition. There’s also an impression that Max For Live is getting some love, as devices like Granulator III look and feel polished and ‘Live‑native’, and Max For Live can now do proper modulation. The only obvious shortcoming is lack of scale and direct tuning support in Max For Live, but I’m sure that will follow. All in all, recommended, perhaps as a late Christmas or New Year present for your studio.