Professional Analog Mastering.
Professional Analog Mastering.
230K+ subscribers
A few years ago, I got bored and tried to reverse engineer the Rode NT1 mic.
I got close, but always had this strange, persistent resonance at 14kHz that I could never figure out, nor could the person helping me with the project.
Since I wanted to use it for at least one project, I tried Soothe 2 but could still hear it. EQ created a noticeable change to the timbre, and any other plugin I tried had the same effect.
That’s when I started spending more time with Izotope’s RX platform, and realized it’s a not so practical but incredibly clean-sounding resonance reducer.
I could highlight the spike at 14kHz, make the filter as accurate as possible, then delete the info - all while having next to no impact on the surrounding frequencies.
If I wanted to, I could copy information from 11-13kHz, and paste it over the area to “patch” the empty spot, so to speak.
Doing this for a resonance that moves around would be way too time-consuming to make it worth it; however, after testing RX against various EQs and linear phase filters, this FFT processor was by far the least destructive.
So, if you’re trying to reduce a static resonance, maybe from faulty gear or a unique characteristic of an instrument, then this is by far the best option.
To show this, I’ll recreate the sound of my knock-off NT1 by adding a 14kHz sine wave to a recording.
I’ll try EQ, Soothe, and finally RX to remove the resonance. Notice how RX is by far the least destructive and the most accurate.
Watch the video to learn more >
More times than not, when reducing resonances, there are only 2 or 3 frequencies that are causing the issue.
Instead of using a full bandwidth of resonance reduction, if we observe the response, we can pinpoint the exact frequency or frequencies to attenuate.
Say we’re listening to a vocal, and when the performer hits 2 notes in particular, we hear the resonance we want less of. These notes could be back-to-back, they could be spread apart, it really doesn’t matter.
The important part is making a mental note of when they occur in the section we’re reviewing.
Once we know ‘hey, at this point I hear the resonance,’ we can listen, observe, and find the frequency or frequencies responsible for any unpleasantness.
Then, we just have to attenuate them with bell filters. Sometimes only a couple of dB is enough to create the needed balance.
You might be thinking, ‘well, I don’t want those frequencies to be attenuated all the time - this is why I need something like Soothe 2 that adjusts throughout the performance.’
But if the resonance is below 4kHz, and still relates to the fundamental of the note, then the resonance likely only occurs when a particular note is sung. In other words, if that note isn’t sung, there’s nothing in that range to attenuate.
So you’re not cutting out anything unrelated - there’s likely nothing there to attenuate until the resonance occurs.
And if you are concerned about this, use a dynamic filter that only triggers when the resonance is present and aggressive enough to trigger it.
This may seem like extra work when you could just use a resonance reduction plugin, but that’s kind of the point. Instead of using a quick but inaccurate method, take the time to make a more impactful change without affecting any unrelated frequencies.
Let’s take a listen to an instrument that has a resonance I’d like attenuated - one I’ve purposefully amplified for this demo.
I’ll use an EQ to balance the resonance and compare it to a resonance reducer. Notice how much unrelated info is affected with the resonance reducer, even when I attempt to control its processing as much as possible.
Watch the video to learn more >
If you’re issue isn’t below 4kHz or in the range of melody frequencies, controlling it may be more difficult.
As we covered in the first section of the video, Izotope RX is a great choice if the resonance is static or consistent throughout a take - but not practical for complex situations.
This leaves us with EQ or a resonance reduction plugin to work with.
However, like our hearing, equalizers and plugins that are frequency-specific like Soothe 2, work logarithmically.
Like our hearing, the fidelity is a lot better in low frequencies than in high frequencies.
For example, say I introduce a very narrow bell filter. In the low frequency range, this filter’,, may cover somewhere between 5 to 10Hz.
But as I move it to the high range and zoom in, notice that the same filter covers a much larger range, usually 50-100Hz.
It only makes sense that EQs would mirror how we hear; however, this makes it difficult to make accurate changes, the higher the frequency range.
So, this gives us a few options in my view.
First, we can use EQ in a similar manner to the last chapter, but with narrow filters and ideally linear phase processing to make the filters more accurate.
Alternatively, we can settle for broader filters. Since we’re less sensitive to pitch in the highs, a broad filter centered on the resonance will still work well - it won’t sound too much like unrelated frequencies are attenuated, since it’s difficult to differentiate between them.
Lastly, this is where I think Soothe 2 works the best.
If we use high-quality, linear phase processing mixed with sharp filters, we can accurately attenuate multiple resonances.
Again, we may not be able to pinpoint the exact frequencies we want, but the attenuation doesn’t need to be exact due to our lack of sensitivity to the high range.
Let’s take a listen to EQ and Soothe 2 used on high frequencies. Personally, I rarely find this type of processing needed, but in case this is useful, let’s compare the 2.
Watch the video to learn more >
Although Soothe is the most popular, people often overlook the Gullfoss EQ.
It works in a very similar way, in which a preset frequency response triggers various filters. Whenever the input doesn’t correspond to the preset response, either amplification or attenuation occurs.
Whereas Soothe solely attenuates, Gullfoss uses both attenuation and amplification to balance the response.
The Tame function is particularly useful for this, in which more attenuation is applied. Since you have the option to utilize masking to reduce resonances, since amplification to masked frequencies occurs with the recover function, I believe it results in a more natural sounding reduction of resonances.
It doesn’t offer as much control as Soothe 2, but the introduction of amplification makes a big difference in my opinion.
So, let’s compare the 2 processors handling the same signal. Notice how both take care of resonances, but the Gullfoss EQ has a slightly more natural sound.