Sonible smart:limit by Sound-Guy
smart:limit from sonible
Austrian company sonible have created some excellent audio plug-ins that are “different” from most, and very effective. Earlier this year they upped the ante on EQs with smart:EQ 3. And now they have introduced a new approach to limiting designed for a specific task: satisfying the complicated balancing act of loudness and dynamics now required for broadcast and streaming services.
. . . . . . . .
What is It?
Smart:limit (s:l) is a limiter with some familiar controls and some unique ones intended for the final loudness adjustment of an audio file. The familiar controls include true-peak level (adjustable from 0 dBFS to -6 dBFS), an input gain control adjustable from 0 dB to +24 dB, and attack/release time controls (including an Auto release time). Note that even using a long attack time does not permit instantaneous peaks to exceed the true-peak setting. There is also an Output gain control (which only reduces gain, from 0 dB to -24 dB).
And there is a preset system where you can save all settings and recall them in any other project. So far, all similar to a normal limiter, but s:l has a lot more tricks in its bag. Instead of just A/B comparisons, s:l has eight States which allows you to store multiple parameter settings in a project. Of course in a DAW project all of s:l’s settings, even the States, as well as any automation you use, are saved. Speaking of automation, I found 14 automation lanes – pretty much all the controls can be automated which could be useful to really fine tune a song's level and dynamics.
As with other sonible audio processors, there is a Learn mode which helps optimize settings for a range of musical styles. This is sonible’s forte, the intelligent, automatic adjustment of multiple parameters. The musical styles are selected using Profile settings with 24 genres/sub-genres of music plus a speech profile and general purpose Universal mode. These profiles set nominal bounds for dynamic range appropriate to the genre and the Learn mode tries to hit the dynamics range for the genre profile. In addition, your can import an audio file that has the dynamics and general tone you want to match, and s:l will analyse it in seconds to use as the genre guide.
. . . . . . . .
. . . . . . . .Profile selection drop-down window with Classic Rock selected.
There is also control of the program delivery system selected using “Publishing Targets” where you can choose streaming services such as Spotify, Sound Cloud, YouTube, Tidal, or Amazon, or broadcast standards such as EBU R128, ATSC Broadcast, or HBO requirements. The Publishing Target defines the maximum allowed true peak level and the expected loudness for a track played on the respective platform.
That’s the basic set of controls, and at it’s simplest you pick a genre that fits your music, select the Publishing Target, play your music and click the Learn button. Smart:limit will set appropriate levels for gain, style, saturation, balance and bass control. You can change the Publishing Target at any time since it only adjusts the expected loudness level, but the genre should be selected before Learn is used since it also affects the five parameters just mentioned above (I’ll describe them later).
Why Do We Want to Do This?
If you are not yet familiar with loudness standards and the methods used by streaming and broadcast services, you should get on board if you plan to create audio content for the Internet, radio or television. If you only record and/or mix, and leave mastering to someone else, you may be excused from this review since smart:limit is a mastering/post-mastering tool. Or you can read on for some useful information.
Perceived loudness of human hearing is a complex issue with many studies over many years going back to Harvey Fletcher and Wilden A. Munson and their 1933 paper entitled "Loudness, its definition, measurement and calculation". Yes girls and boys, long before digital audio, DAWs and the Internet. For a long time broadcasting and even the Internet simply played audio so that the peak level was no more than a specified level, 0 dBFS for digital audio. This helped generate the”Loudness War” that plagued music for decades. In the last decade, audio engineers have sought to better control loudness, leading to various international standards such as the EBU R128 standard that defines methods to measure loudness, and the now common metric, LUFS (Loudness Units relative to Full Scale).
Unlike peak measurements where two songs with the same peak level can have extremely different perceived loudness, any music with the same LUFS level will have nearly identical perceived loudness. And the various streaming platforms (Spotify, YouTube,Tidal, etc.) and broadcast services now impose specific LUFS targets, typically in the -14 LUFS range for streaming and even lower for broadcasting (in the range of -24 LUFS). Note that streaming services may change their target loudness levels in the future, so you might want to confirm that a given platform is still using the levels suggested by s:l. You can easily set the reference loudness parameter to a custom value, and sonible will no doubt provide updates if new platforms become available or significant changes are made to the current platform levels.
During the ‘loudness war’, music was sometimes pushed to what now would be measured as -4 or even -3 LUFS by compressing or limiting the dynamic range of the audio material so that the peak level was still at 0 dBFS. This made for some real crap music, even though it sounded much louder than a properly mixed/mastered song. Today, compressing a song to only 4 dB of dynamic range buys you nothing since it will be streamed at something like -14 LUFS leaving 10 dB of empty headroom with no audio benefit!
How does dynamic range fit in? This is a concept that confuses many people, including some otherwise experienced audio engineers. And in fact, it is a measurement still being evolved, so some confusion is understandable. Basically, the EBU R128 and related standards define dynamic range as the difference between the true-peak level and loudness (in LUFS). So audio with a loudness of -14 LUFS and a true-peak (instantaneous maximum value) of -2 dB has a dynamic range of 12 dB. At least that’s the basic definition, but it gets a bit complicated in practice. LUFS measurements require some rather complex time and frequency filtering, so they are not identical to dBFS measures that some DAWs show, although many DAWs today provide a LUFS measure. And there are three defined LUFS metrics, integrated (over the full length of the audio program), short term (a 3 second moving average), and momentary (a 400 msec moving average).
Luckily for you, sonible have done all the maths in s:l, and while they were at it, they created a new, improved dynamics metric! If you are familiar with the loudness standards, you will no doubt be surprised that the s:l dynamics numbers are not the same an an accurately calibrated LUFS meter (such as Klangfreund’s excellent Multimeter). When I first tested s:l I thought something was wrong since the Dynamics reading was not meeting the EBU R128 standard – it was always higher by about 1 to 3 dB. Sonible have developed a metric they feel better matches the perceived dynamics of human hearing – they calculate the median value of all measured PSR values (true-peak to short term loudness) rather than just using the simple difference between true-peak and program loudness (integrated LUFS). They point out that PLR, the dynamics measure defined in EBU R128, is often misleading, not representing dynamics as we hear it. And if a track is normalised, for example to -1dB true-peak, PLR is just another descriptor for the integrated loudness.
And There is More
The style, saturation, balance and bass control settings, mentioned back near the start of this review, are what sonible call the sound-shaping panel. They provide some fine tuning of the limiting mode, saturation (added harmonic distortion – odd harmonics from none to about 3%), spectral balance, and prominence of bass frequencies (basically a bass-boost). The Learn mode along with your audio file and the selected genre will suggest settings, but you are free to change them as you see (hear!) fit. You can also select the LUFS response time as described earlier (integrated, short or momentary), and there are some other useful controls including constant gain that removes loudness bias when you switch between s/l processing and the unprocessed audio, a channel link control that let’s you adjust the amount of linking between channels when working on stereo or multichannel signals (up to six audio streams are supported if your DAW can handle it), and a delta mode that let’s you hear what s/l is “removing” from the input signal. Since s:l is a limiter and not a compressor, the delta signal can be quite “snap, crackle and pop” sounding, but that’s what a limiter does!
There is a Quality Check switch near the bottom of the GUI to the right of the sound-shaping panel that can guide you in fine tuning parameters to get your track ready for publishing. While enabled, the Quality Check continuously analyses the dynamics and loudness of your audio signal, compares the results to the ranges suggested by the genre profile (dynamics) and the Publishing Target (loudness), and makes suggestions for improvement. Note that the constant gain switch should be used after Quality Check is run – if it is left on, Quality Check will never give you a suggestion.
. . . . . . . .
. . . . . . . .Quality Check engaged showing low dynamic range (8.6 dB in red at the upper right) and noting that the chosen platform
. . . . . . . .(Spotify) will turn down the track loudness by 6.6 dB (lower right pop-up window).
The readouts in the upper right include LUFS (using the selected integration time), Loudness Range, Dynamics, and TP (true-peak). Dynamics and true-peak have already been discussed, but Loudness Range is the variation of short term loudness based on the loudest and quietest non-silent sections within the measurement period. You typically will want at least few dB of variation in a song since verse, chorus and bridges can have more impact if they are not all identically loud.
s/l has an excellent graphic plot, the Loudness and Dynamics Grid, that I find very informative. As you can see, this shows loudness on the vertical scale and dynamics on the horizontal scale (with lower dynamics levels to the right, and higher ones to the left, which feels a little odd, but works OK once you accept it). This plot has a number of features including an excellent real-time indicator (a small x) that moves around the plot as momentary loudness and dynamic range changes throughout a song. And a larger + symbol slowly settles on a loudness/dynamics value as the music plays, displaying the integrated loudness and the sonible overall Dynamics value. There are also plots of the loudness and dynamics distributions shown along the loudness and dynamics axis. And another informative plot is available at the lower left, the Distortion Monitoring display which indicates the amount of distortion in bright red across the frequency range. The user manual describes all this well if you want more details.
. . . . . . . .
. . . . . . . .Distortion Monitor turned on and showing very high levels across the full frequency range (lower left) with the Loudness
. . . . . . . .and Dynamics Grid on the right showing a distribution of the low dynamic range at the bottom.
A fascinating and useful tool for rapidly setting limiter parameters to provide compliance to a range of audio delivery platforms with additional sound-shaping controls to enhance the resulting audio in several ways. I found that unless I manually changed some settings to the extreme, the results were transparent, and could even be enhanced by using the sound-shaping panel. Definitely a tool to try if you are producing music for broadcast or streaming services.
smart:limit is available as Audio Unit, AAX, VST2 and VST3 – for 64 bit OS only. Windows 10 and Mac OS X 10.9 or later are officially supported. Requires iLok dongle or online PACE system connection (free) for activation, but runs without a connection after authorization (though the trial period requires a connection when you open an instance). Supported sample rates are 44.1 kHz to 192 kHz. In my test system (PC Audio Labs Rok Box PC with 4-Core Intel i7-4770K, 3.5 GHz, and 16 GB RAM) a single instance used from 0.8% to about 1% CPU resource depending on the settings. Latency is an odd 2,704 samples at 48 kHz sample rate (about a 60 msec delay), but since smart:limit is a mastering tool (even a post-mastering tool) latency should not be an issue.
Unique Learning mode that automatically “tunes” limiter parameters to best match an audio file to musical genre and streaming or broadcast loudness requirements.
Over two dozen musical genre profiles available and twenty-two playback platform requirements that help control and monitor loudness, dynamic range and peak signal level.
Easy and quick adjustability of the controls with real-time feedback on the excellent Loudness and Dynamics Grid.
Unique sound-shaping panel enables adding audio enhancement while monitoring the loudness and dynamics.
Excellent sound quality with very transparent control of both loudness and dynamics.
Continuous GUI size from 600 x 1,000 to 900 x 1,500 pixels.
Free 30 day trial with full functionality.
Since smart:limit is unique and does an excellent job with its intended task, I can’t complain. I would love to see the Loudness and Dynamics Grid available as a standalone plug-in since I can see it could be sometimes helpful in a mixing situation.