FORUMS: list search recent posts

Audio limiter choice & settings

COW Forums : Adobe Premiere Pro

<< PREVIOUS   •   FAQ   •   VIEW ALL   •   PRINT   •   NEXT >>
Alexandre Brandt
Audio limiter choice & settings
on Mar 6, 2017 at 9:23:43 am

Hello all,

I've got a short track of music in sequence. The audio dynamic is fairly wide, between -12 and -2.
I'd like to limit it at -6 for as it's for a web video.

I tried a few audio effects that come with Premier, the limiter, the Mastering the compressors but I can't manage to get working properly, getting a decent setting just to limit at -6 with the least distortion possible.

Any advice ?


Producer | Filmmaker
Geneva - Switzerland

Return to posts index

Dave LaRonde
Re: Audio limiter choice & settings
on Mar 6, 2017 at 3:22:28 pm

Do this in Audition, with better audio tools.

Dave LaRonde
Promotion Producer
KGAN (CBS) & KFXA (Fox) Cedar Rapids, IA

Return to posts index

Alexandre Brandt
Re: Audio limiter choice & settings
on Mar 9, 2017 at 2:12:54 pm

Ok thanks.

Producer | Filmmaker
Geneva - Switzerland

Return to posts index

Simon Billington
Re: Audio limiter choice & settings
on Mar 16, 2017 at 2:11:06 am

Limiting a huge dynamic range can prove to be problematic, as you have discovered, and can often result in unwanted artefacts. It's not necessarily about the quality of the processor, more often than not it's a simple matter of physics and you may be asking more from your processor than it is capable of doing without unwanted side-effects.

Generally I'd suggest compressing a track as well, but you can only do so much of that before it brings up noise on the dialogue track. So more often than not you require a chain of several processes to get something close to the desired effect.

A chain such as this one...

  • Automate Volumes

  • Soft-Clip

  • Compression

  • Equalization

  • Limiting

  • You don't have to do do all of that, but its the sum of all the parts that's what get's you're audio to sound better. Just a bit of each, its the sum off all the parts.

    Automation can be tedious and a pain in the ass, but you just want to focus on the big moves to get all the audio roughly equal loudness. Use a loudness or RMS meter and target for somewhere between -18 to -23 dBFS. Choose a level and stick to it as closely as you can while just focusing on broad level changes to keep everything in check.

    There are gain riding plugins out there that can do this much more efficiently and with more detail if you want to check them out. I use this one here

    You can set a normal compressor up to behave in a "soft-clip" or "soft-limit" kind of way. The idea here is that it compresses the peaks, similar to a limiter, but it does it more transparently, to a certain point. You don't want to abuse it, just use it to rein in some of those extremes to make them more manageable. You could also use a soft-clipper style plugin to make life simpler, you just have less control. IK Multimedia have one.

    If you choose to use a compressor to do the same job it needs to be able to do ratios lower than 2:1 and have a minimum attack of around 1ms or lower, as well as have an ultra-fast release, anywhere under 30ms. After setting the attack and release and adjusting the ratio to around 1.5:1 give or take, lower the threshold till you start seeing a tiny bit of action on the metering.

    Bypass the plugin to compare, the effect should be barely noticeable. If you can't hear the effect try applying a bit more, by lowering the threshold even further. Apply as much as you can when the effect becomes too noticeable, back it off. The idea is you want it to be transparent but reduces as much of those peaks as you can. It's bit of a balancing act. You could get as much as 6 to 10dB of gain reduction before the effects become too obvious, depending on the audio and how well the compressor is set up. But watch you don't increase your noise levels much.

    If the audio appears to be a bit dark, or not as bright sounding, try reducing the release, no less than 5ms or increasing the attack, not past 3ms or a bit of both. Generally because you want to iron out all the attacks though, I would leave the attack alone if possible. If you incur a bit of distortion your release may be too short.

    Keep bypassing the compressor and checking against the unprocessed audio, to make sure that you are not going overboard. However, many compressors have a mix or blend knob and you can always pull down the effect a little if you think you have gone overboard. Otherwise try just raising the threshold a little to reduce the strength of the effect.

    This time you are wanting to set up a compressor with a slower attack and a relatively fast release, as well as a higher threshold and ratio. Almost the opposite of what you did with the clipper. You are only looking for around 1-3dB of gain reduction though, or thereabouts. Much more than that and it could bring up the noise level on the track. However, if you run a good denoiser on your audio you will find that you would be capable of teasing more level out of the track before the noise levels get too high. There will be some trade off, though. You are bound to increase the noise at least a little. If you find it becoming too distracting then back off the compression.

    In the case of this compressor, the attack would want to be around 5-30ms and the release around 50 to 150ms. Set the Output or MakeUp gain to compensate for how much you wish to compress (1-3dB), set the ratio to about 2:1 to 3:1 and lower the threshold until you get the desired reduction.

    If it sounds a bit "suffocated" try increasing the attack, or if the meter isnt recovering quickly enough between major "transient" events, reduce the release. If that doesn't work, reduce the effect by lifting the threshold.

    As always, bypass the process and compare. The processed version should be roughly equal in loudness to the unprocessed sound, if not adjust the Output or Makeup gain. It should sound more energised at this stage, if not lower the threshold a bit and compensate for the overall change in level. If you want the effect to sound "snappier, increase the ratio a little and maybe the attack, if it's too snappy, or percussive turn down the ratio, or back off on the threshold.

    Next you may want a little eq just to brighten things up a little or add a bit of low-end presence. 8k to10k boosts can help lift the audio and give it a bit of air. A boost around 150-250Hz can give the dialogue some weight. A high-pass filter rolled off under 80Hz could reduce roominess and some plosives.

    900Hz to 1.5K can be pulled with a narrow Q, if the dialogue is to nasal, or boosted to add a little more cut, too much will make it go nasal though, obviously. Too much of 3-5kHz can also make audio sound harsh, but if it's sounding dull, you may want to consider boosting somewhere in this area. Anywhere from 300-600Hz can be used to reduce a muddy, "roomy" or boxy sound, if you're dialogue sounds thin, however, you may want to consider adding some back in.

    If you are eq'ing a mix of dialogue, music and even effects, then you will want to consider using only small adjustments, less than 3dB, and you are looking for a good overall balance between everything.

    FINALLY we get to limiting!!

    Get as much as you can out of this guy before it starts sounding distorted or too percussive. By this time you should be able to hopefully raise your audio to around -13LUFS for YouTube, or whatever standard you are working with, and get a much better level out of your mix. Doing everything up to this point will REALLY HELP in boosting your final levels and getting more satisfactory results.

    As i said earlier, its a sum of all the parts.

    It takes some time to develop an ear for compression, but you have to start from somewhere. You will start to find that all the controls are co-dependant and adjusting the one can effect the way the others respond. I have given you a good ball park place to start, so hopefully it will be much harder to get it wrong, other than overdoing it, but not impossible.

    I suggest that after setting it all up, walk away, comeback after your ears have had a break and listen to it again and see if you have gone to war with it, or whether it requires more. Turn off all the effects one by one and then slowly add them back in, making any necessary adjustments, but with fresh ears.. Be aware that tiny changes early on in the effect chain can have a more drastic effect on other processors down the line.

    It's possible to find multi effect units that bundle most of these effects in the one plugin. The D5 is a good example, except it misses out on the eq, you can always add that after the fact though. On the plus side, it is a quite clean type of processing and also light on CPU resources. Then there are other tools like Neutron or Ozone, but they are quite expensive for most to consider.

    I am really dreadfully sorry this turned out to be an epic post!!

    I'm hoping someone benefits from it, if not yourself.

    Return to posts index

    << PREVIOUS   •   VIEW ALL   •   PRINT   •   NEXT >>
    © 2018 All Rights Reserved