MeldaProduction MSpectralDynamics by Sound-Guy
MSpectralDynamics and new MSpectralDynamicsLE by MeldaProduction
I must admit to having many plug-ins and having tested a lot more than I still have, and yet up to now I had tried no processors from Czech company MeldaProduction. Considering they have over three dozen free plug-ins and more than double that number of paid processors, and they get fine reviews, it’s a bit odd that the new MSpectralDynamicsLE is the first MeldaProduction product I’ve tried!
What is it?
MeldaProduction make some unique audio processors that don’t look or work like most plug-ins and MSpectralDynamics and the new LE version are great examples. MSpectralDynamics has been available for a few years and has received good reviews with one aspect that several have complained about. It is not easy to understand or use without considerable effort. The reason the full version is a challenge is because there are so many functions and features! The LE version simplifies operation while providing the main functions of the full version – in fact, the LE version actually does exactly what the full version does in its “Easy” mode, but has no access to the full Edit mode.
Easy Mode screen with eight basic functions (or “devices” as MeldaProduction call them).
Edit Mode set for a “flattening” process. Note the extra controls on the left and even more settings on the right.
MSpectralDynamics is a dynamics processor that works in the frequency domain rather than the time domain, although it also performs some functions in the time domain. This enables it to work with individual frequencies, providing compression (or expansion) that no regular compressor can. In one sense this is like a multiband compressor with thousands of bands – one that can use a sidechain signal for controlling gain reduction. And it can do more than audio compression.
Spectral ducking mode.
The image above shows a spectral plot of pink noise on the left, which would be a flat line (using a slope of 3.0 in Voxengo’s SPAN), but a sidechain signal was used here (a vocal track) and the gain reduction occurs proportionally to the levels and at the frequencies of the vocal in real-time. Well, not actually “real” time since such complex processing requires some latency – in this case 8,192 samples (a 171 msec delay). But your DAW's PDC will keeping everything in line. On the right is the MSpectralDynamics “Easy” view, essentially the LE version. Note the eight functions listed: Flattening, Compressor, DeEsser, DeNoiser, Ducking, Ducking Bass & Drum, Gate, and Spectral Crusher. These are very useful and basically all that the LE version supplies – the full version can directly access these using its Easy view, but has a lot more capability in its Edit mode. We’ll start with Easy.
Easy Does It
The user manual has some issues. It is confusing in a number of places although it does cover pretty much everything. The first few pages refer to the full Edit mode but show only the Easy view which threw me when I first read it. The manual is packed with information on the functions and controls and there are plenty, but not always clearly described. Considering the Easy mode first, it includes the eight functions I listed (MeldaProduction call these “devices”), and a small number of controls that change to fit the device used:
Four of the Easy “devices” which are available in the LE and full versions.
In each case a control in Easy mode usually affects several detailed adjustments that can be separately adjusted in the Edit mode. In the Easy mode these interact to provide a useful range of processing.
Flattening is a spectral compression mode that applies modest compression (~1.25:1 ratio) over the full dynamic range of the input signal (threshold is set very low, -80 dBFS). It uses a fast attack and release, and it applies gain reduction based on the instantaneous input level at each specific frequency. Nothing like a normal compressor.
Compression is gain reduction based on input signal level, but can also control specific frequencies and the “focus” can be changed from very high Q (narrow band) processing all the way up to wideband. Like Flattening, the gain reduction is based on the input level at each specific frequency, but the threshold is in the middle of the level range (-40 dB or so), the ratio higher (1.5:1 or more), and attack and release times are set to moderate levels. Again, it does not operate like a normal compressor!
DeEssing is another frequency selective gain reduction process, but with the operating frequencies limited to upper middle to high (3 kHz and up). Since the “ess” sounds are usually smeared across the high frequency range, MSpectralDynamics doesn’t need to focus on individual frequencies as it can with spectral compression and flattening. It uses a very fast attack and moderate release time, a high threshold (-20 dBFS), and a limiter mode (oo:1 ratio).
The DeNoiser works like an old standard finite impulse response (FIR) processor I’ve used for years – but does a much better job than it can do. In fact, I was impressed how transparent the noise reduction was even when I used test signals with the snr only 20 -15 dB. It works even on full mixes, but better on single instrument or vocal tracks if they have objectionable noise floors.
The two ducking modes vary mostly by the frequency range limitation applied with the drum & bass mode that limits action in the 100 Hz to 1 kHz range. These modes use a sidechain input and apply the spectral processing so that the gain reduction of the ducked channel is applied only to the exact frequencies from the control channel (the “important” sound you want to focus on).
The Gate uses a dedicated gate module (accessible directly in the Edit mode) that provides a threshold and some spectral controls (it’s not your grandfather’s gate!).
The Spectral Crusher uses a very different compressor transfer function – a stepped mode – that can provide some really odd distortion FX.
The Full Deal – the Edit Mode
If you use MSpectralDynamicsLE, the above eight “devices” as Melda call them are what you get and are very effective and useful. But if you’re a hard-core techie, the full version includes the Edit window that opens up an astounding range functions. The Edit button found in the top menu bar switches between Edit and Easy views and a very useful feature is that if you start in the Easy mode and choose a "device", then click Edit, you will see all the details of the settings that make up the "device". I found it a good way to learn how the detailed controls actually work. You can jump back and forth, changing settings in the Easy mode and see what happens in the Edit mode. But going the opposite direction doesn't sync – changes made in the Edit mode are not reflected if you jump back to the Easy mode. This makes sense because there are many adjustments in the Edit mode that cannot be controlled in the Easy mode.
Edit includes access to spectral properties (resolution, smoothness, naturality), detector selection (input, sidechain, input+sidechain, input-sidechain, sidechain-input, multiply, minimum and maximum), selection of non-linear transfer shapes, dynamic detection parameter adjustments including custom attack and release envelope shapes, parameter adjustments for two dynamics processors, a very comprehensive modulation module which lets you control parameters using various inputs, and extensive MIDI control functions.
The two dynamics processors each can provide compression and expansion – in fact they can provide downward compression, upward compression, downward expansion and upward expansion! If you’re not familiar with these variations, downward compression reduces gain above the threshold setting while quiet sounds remain unaffected – it’s the “normal” kind of compression. Upward compression increases the gain below the threshold while leaving louder sounds unaffected. Both modes reduce the loudness range. Expansion is the opposite: upward expansion increases gain above the threshold setting while quiet sounds remain unaffected. This mode can quickly push the output level above 0 dB. MSpectralDynamics has a built-in limiter that can be switched on to prevent output levels over 0 dB since driving a high level into subsequent plug-ins will likely cause distortion and possibly bother your neighbors! At any rate this mode must be used with caution. Downward expansion reduces gain below the threshold which is like a gate although simple gates cut off the sound below the threshold setting rather than just reducing gain.
Note that compression is applied when the Ratio setting is “greater than one” (ie, 1.5:1 or 3.0:1) and expansion is applied when it is “less than one” (1:1.1 or 1:1.5). The downward button is just left of the Enable switch.
And since this is MSpectralDynamics, the gain changes are not just applied evenly across the audio spectrum – depending on the setting of Smoothness, they can be focused tightly on “each” frequency (low Smoothness) or evened out (high Smoothness) so that the results are more like a normal compressor.
The Edit mode has a lot you can do, but I won’t try to describe more here since it would take too much time (mine and yours) – and I haven’t tried it all myself yet! If you are "hard-core" tech oriented, you’ll be able to spend hours, days or more exploring what is available and can produce a large range of dynamic and spectral effects. Definitely not your father’s toolbox!
Both versions have factory presets and enable user presets to be saved, both provide Undo and Redo, and a numeric keypad that pops up when any dial or slider is double-clicked for exact entry of a setting. Touch screen displays are supported on Windows 8 and newer Windows OS with support of up to 16 connections/fingers/inputs.
MSpectralDynamics is available for VST, VST3, AU and AAX interfaces for Windows and Mac OS. The installer automatically installs both the 32-bit and 64-bit versions of the plugins. In my test system (PC Audio Labs Rok Box PC with Windows 7, 64 Bit, 4-Core Intel i7-4770K, 3.5 GHz, and 16 GB RAM) MSpectralDynamics required from less than 0.5% to a little over 1% CPU resource depending on settings. Latency runs from 1,024 samples at the Low Quality setting to a whopping 32,768 samples for Extreme Quality, a quality level I see no need to use myself. I used High Quality in all my testing which has an 8,192 sample latency, 171 msec, so this is not a processor for live performances!
“Wow!” hardly covers it! This is a serious set of audio processing tools and even the LE version will provide excellent capabilities that are not commonly available. The spectral ducking modes, denoiser, and flattening functions alone are worth the price of the LE version, but if you like to spend time creating original processing modes, the full version is something special. And will keep you busy for months. Note I gave "Ease of use" five stars which the LE version certainly deserves – one might argue the full version should get four (or less?) stars for "Ease of use", but I can only use one rating, and if you are brave and persistent you will find the full version worth every effort to master!
Excellent spectral (frequency domain) processing that provides some unique capabilities.
LE version does a lot with minimum user inputs.
A Swiss Army knife of audio processing tools.
Can handle stereo, Mid-Sides processing and surround sound up to 8 channels.
There are 12 visual styles for the GUI – I used the default "2D" Neon view here, but there are some "3D" styles and different color schemes if you like.
Full version can do things with audio I’ve never heard or thought of before – an audio nerd’s wonderland.
And until October 10, 2020, these plug-ins are on sale.
The full version is extremely complex and can challenge even experienced audio technologists.
The user manual is confusing in places, though it provides good information on most of its 115 pages. More clearly explained examples would be beneficial.
https://www.meldaproduction.com/MSpectralDynamics and https://www.meldaproduction.com/MSpectralDynamicsLE