Hi,
is there any way to negatively increase the Amount of db in de Limit-Filter to -30?
The maximum amount which can be set seema to be -20. But maybe there is a trick or another filter which can help or even an extension for this.
Could you add the filter twice?
Add the filter twice?
Is this a joke or a real suggestion?
@brian Like I have already written this seemed to be a joke but I wanted to be sure.
Applied the “Limiter” Filter twice to an audio channel and yeah what should I say… If I have two Limiter-Filters on the Audio-Channel and both are at -20 db then the maximum Limitation of the Audio is… tada -20 db.
But I wanted to have a Limitation of -40 in this example. So maybe it should be possible to set a greater value to maybe -40 or even -50.
It wasn’t meant as a joke. You asked if there was a trick and I thought that might work for you.
I do not understand the use case for such high degree of limiting. I think the current range is good for normal uses.
Why I thought this was meant as a joke is that this doesn’t changed anything at all. And my understanding is that the filter “cuts” everything “after” or above the Set Maximum.
So if I set a Limit(er) for -20 db than the Sound which would peak above the -20 db will be lowered to -20 db. So this is a maximum.
And if I now add a second Limiter then this Limiter would just do the same. But not from the previous Set of -20 but from 0 db.
Right?
So how should this then help if I add a second Limiter?
And yes I would definitely need a higher range. So I have Videos which range from silent to around -20 or -15 db after normalization. And these consist of game sound of Lets Plays and recorded Mic-Sound. The Game sound is around -20/-15 db and the Mic in a max around -25 db. So I would now like to maximize the Game Sound to around -30 db so its always lower in loudness than the Mic. And one can always clearly hear my Voice. And the Limiter would definitely help me with this.
And the plain Loudness-Filter doesn’t help because the Loudness of the Game is not really predictable. So I would preferred go with the Maximum of db with the Limiter.
I see what you mean now. I guess I was not thinking critically when I made my suggestion.
Understanding your use case is helpful. Based on your goals, I would not recommend to use the limiter filter. I think it would result in a lot of distortion (especially if you let it target such a low level).
In general, here is how I would describe the various filters:
Limiter: Ensure that peaks to not exceed a defined threshold. This is usually a blunt limiting and can result in distortion if applied to aggressively. Think of this as “Break glass in case of emergency”.
Compressor: Reduce the dynamic range of a clip so that it does not exceed a maximum. This usually has a “softer” approach to the gain reduction to reduce any perceived distortion.
Normalize (1 pass): Target a specific loudness level by averaging the overall loudness. This will both decrease and increase the sound level to achieve the target loudness. Also, this has a range that is within your target level.
Based on your description, I would suggest to use the Normalize: One Pass filter. If you find that the filter does not respond to loud moments fast enough, you could put a compressor in front of it to smooth out the loudest section.
I’m no expert, but what about adding a Gain/Volume filter (below the Limiter filter in the filters list) to remove the extra -10 dB ?
No joking here, it just seems to work well in the test I made.
1 Like
Yes. Absolutely. Actually, if your background music is a consistent volume, then all you should need is the Gain/Volume filter.
Okay thanks.
But I still dont get why this value is limited. Why isn’t this just a text field and I cant enter whatever value I want? If it results in bad sound its my fault isnt it? Why is there a discussion about such a thing which shluld be a smal change?
For me right now its a bad situation because I cant really control the Sound to a level I want and I would need to hack something. Which results in even worse results.
The audio processing code itself has a -20 dB limitation. A simple GUI change would not be sufficient. It would require a change to the math in the limiter code itself. Doable, but it takes more time than you’re hoping. See:
https://mltframework.org/plugins/FilterLadspa-1913/
As others have said, there are better ways to tackle this problem than a -30 dB limiter.
The first priority is to get the music and other background tracks into a small dynamic range, so that soft sounds don’t fall entirely out of hearing range. Using a Compressor is a good first step, although it is usually not necessary for music or sound effects that are professionally produced. Those sounds are already well-contained.
The next step is getting the average perceived sound level to -30 dB. The Normalize: 1-Pass filter should be sufficient for this if the dynamic range was contained as described above. If not, the Nornalize: 2-Pass filter can look ahead to do a better job. The Limiter is not a good tool because it looks at peak absolute values instead of average human-perception values. If the goal is to have a consistent-sounding volume level for your listeners, then the sound needs to hold steady at a perceived average level, measured in LUFS, which the Normalize filters provide. The Limiter does not. It just crushes any sound that peaks without doing anything to hold the meat of the sound steady. The Limiter does not consider or compensate for different songs having very different distances between their average volumes and their peak volumes. This is why it is a poor choice for volume management, and is best used as a safety tool during mastering… unless you’re recording distorted guitars and scream metal.
The last step is to add an EQ filter that notches down the frequencies of your voice in the music tracks. Play a clip of just your voice, watch the frequency spectrum, and note where your voice peaks the most. Notch those frequencies down in the music track about -3 to -6 dB to create a hole in the music for your voice to sail through. This is the real key to making your voice clear… eliminate competition in other tracks.
It is entirely okay if the music rises above -30 dB on occasion. It’s the perceived average that matters, along with the way it combines with your voice. If the music track has the EQ notches, the music can rise as much as it wants and still not drown you out because there is nothing in that range to compete with you.
1 Like
If it’s still relevant. The most efficient way to limit the audio to -30db would be like this:
-
- compressor
-
- limiter
-
- gain -30.
The compressor is needed to make very quiet sounds louder, and make loud ones quieter, that is, to reduce the dynamic range, which is very useful for voices, for example. The main thing is to look at the “gain reduction” parameter so that the sound does not compress more than 6-10 dB, otherwise there will be a clamped brick. Either the “makeup gain” of the compressor, or the input gain of the limiter is raised until the sound peaks at 0db. And after adjusting the compressor and limiter, you can lower the overall volume by -30 dB (although as far as I know, TV and RADIO require -23, but not always).
What benefit does the limiter add that the compressor didn’t already provide? Wouldn’t the -30 dB gain be so drastic that no limiting was needed at all if gain was done first?
Granted, I still find value in a -1.5 dB limiter on the Output track, but I didn’t know if that is what you were suggesting here.
European television uses -23 LUFS, not -23 dBFS. USA television anchors the dialog track to -24 LUFS. Aligning peaks to -0 dB doesn’t tell us anything about the average volume. If we align (normalize) peaks of a symphony orchestra to -0 dB and also a rock song to -0 dB, the rock song will sound far louder because it never gets as soft as the orchestra. The dynamic range of the rock song is smaller, so the average volume will seem louder. This is why adding a gain of -30 dB will not provide consistent results on different music types. Using LUFS accounts for this, because LUFS is the average volume weighted for human perception. This should allow the OP to speed up their workflow because -30 LUFS on the Normalize: 1-Pass filter results in roughly the same volume regardless of what audio is being pumped through it. No need to fiddle with the Gain filter to find a custom value that works with each song.
The limiter is the compressor, whose “ratio” parameter is always equal to infinity. The limiter is a brick wall that will severely limit the sound, and is mainly used to deal with peaks. The compressor that is in shotcut is limited to a ratio of 1:20, which is good compression, even overkill, but it will still somehow skip peaks above the set limit, especially if RMS is prioritized. To get a good result, I set a compressor on the tracks with sound, and a limiter on the master channel. As a result, I get exactly the sound that I need.
p.s. By the way, of all the free video editors, shotcut has the best and most easy-to-use compressor. In other programs, either it is not there, or it behaves inappropriately. For shotcut, it works like a normal compressor should.
1 Like
That’s the detail I was looking for. As opposed to a limiter on each individual music channel. I wasn’t seeing any benefit to limiting each clip individually after 30 dB of reduction, when a single limiter on the master would suffice.
1 Like
This topic was automatically closed after 90 days. New replies are no longer allowed.