Peaky notes, EQ and compressor limitations, uncontrollable

‘lowering notes by decibel and hertz’

Seems the EQ and compressor do not discriminate when reducing and lately realizing only certain parts of a song have a frequency too loud (I want to squash!) – I’m thinking like a gate type EQ or compressor that will only lower parts louder then what it’s set at. For example parts of song too loud at 200 hertz but not wanting to lower the 200 hertz at other parts not too loud as it seems the EQ and compressor does. I will be trying some other compressor and EQ plugins asap but at this point I’m still just using the default Audacity ones. And having to meticulously edit per spot by ear and plot spectrum, then EQ those wild parts down. Not really even needing compressor when EQ is precise it seems…

So is there… an EQ or compressor with such capability and what is the terminology used describing such and such features if they exist? thanks.

Assuming an EQ when set at say -3 dB at 200 hertz goes through and lowers all 200 hertz -3. When what I am really wanting is say lowering all 200 hertz above say -18 dB down to -18 dB, as an example.

UPDATE: I’ve switched sites around and at this time have a new mostly proprietary method of things (previous voided to Apr.19 2017). If you like my new sound and want to know more about my editing techniques feel free to ask,

Ronald Newman
http://www.SoundCloud.com/BlackDogSongs

The closest is a multiband compressor…

A free one [which works with Audacity on Windows] is available here … GVST - GMulti

I think you’re over-analyzing this. Volume spikes respond very well to either Effect > Compressor or Effect > Limiter. Either of those tools will reduce volume spikes with only very slight effect on the show. In most cases, you don’t notice they’re working.

Ripping a show apart into frequency components is dangerous because you will be able to hear that. That will change the character or even pitch of an instrument, voice or performance.

Koz

Thanks all, great, I’ll experiment and Koz, you’re probably right. That was more a trial and I agree, did seem overkill to the point of, well, not working? Also, I am working with one stereo track live acoustic guitar and vocal no dub recordings which is a challenge it seems, and probably a big part of the problem, not having separate tracks!

Probably partly why acoustic guitar with vocals is not more prevalent in modern music. There are the old acoustic guitar and vocal blues recordings however… and how they recorded and engineered is no doubt a lost art. Of course digital may not transfer that art much, regardless, interesting stuff I’d like to know about.

I think the tool you’re looking for is a limiter

Audacity's Limiter.png

Thanks Trebor and weird I have been using that, and that one, only recently, that is a great tool. I also went with sc4 on last song [first thing], set for peak usuage, throttling the ratio and or threshhold to do a similar task, bringing down the peaks first thing when editing. I think I maxed the ratio then eased the threshhold on until it sounded good… I’ll have to look at my notes [if that’s not exactly what I did I will try to update this post or thread ASAP if I can remember!]. thanks thanks

BTW only prob is it only goes 10 dB deep, any other similar limiter that goes deeeper you know?

You can apply it repeatedly, two applications will be 20dB , ( if “make-up gain” is on “Yes”, as shown above).
Pressing “Ctrl”+“R” in Audacity repeats the last effect you applied.

Thanks Trebor but seems that would anticipate more distortion, as opposed to just one step… {?]… Well actually I was using on segments as well, not whole song, so I’d have to even try to get the segment back at proper volume after that route… I guess I could raise it -say 10 dB then use the limiter then lower it 10 dB… but better to find a deeper cutting limiter…

Question 1. RMS compression question:

What I did new was at threshold setting… but should the threshold (when doing RMS compression) be set above the peak per plot spectrum analysis (if so, why?)? Which I guess (the plot spectrum value) is the RMS value for the peak [?]. Example: spectrum shows peak at -13 dB so I set the compressors threshold at -12 dB…[?]

Question 2: To the pros, or experienced:

Is there a standard order in which effects should be applied? I.e., EQ, compressor, de-esser, limiter… etc…

Thanks.

RMS is a kind of “average” value. The RMS level of audio is always lower than the peak level (actually, the RMS level of a perfect square wave is exactly the same as the peak level, but that’s the one and only exception).

Audacity shows the RMS level of the waveform as a lighter blue color within the darker blue waveform.
firsttrack000.png
When using compression based on the RMS level, you need to set the threshold below (lower than) the maximum RMS level of the audio.

With those effects, no there is no standard order, though you will get different results depending on the order that you apply them.
The explanation for this is fairly straightforward:

(I’m assuming the limiter is a conventional limiter and is not applying “make-up gain”)

Let’s say that you have an audio track that has a peak level of -6 dB.
Apply a limiter with the threshold at -3 dB. It will have no (or very little) effect.
Now apply EQ with a substantial amount of boost. The EQ is applied and the peak level increases.

Now do it the other way round.
Apply EQ with a substantial amount of boost. The EQ is applied and the peak level increases.
The peak level may now be above -3 dB.
Apply a limiter with the threshold at -3 dB. If the peak level after applying EQ was above -3 dB, the limiter will reduce the peak level to -3 dB.

Thanks all!!

Stumped!

SEE BELOW

Thanks Steve, but this isn’t clear to me. I have two questions:

  1. How do I determine “the maximum RMS level” of an audio with Audacity?

  2. If the maximum RMS level of an audio is -10 dBs would -11 dBs be below this and an example of a correct RMS compression threshold setting?



  • You can estimate by looking at the height of the light blue region of the waveform
  • For mono tracks you can measure the RMS level with “Analyze > Contrast”
  • You can use one of the Nyquist plug-ins such as wavetats.ny

In the next version of Audacity, “Contrast” can handle stereo or mono.

Yes -11 dB is below -10 dB, but only a tiny bit below. As a compression threshold setting, a -1 dB difference would be very subtle, perhaps too subtle to hear.

Three things worth committing to memory about “dB”:

  1. 0 dB = “full scale” (maximum “valid” signal level. Integer formats clip at 0 dB)
  2. -inf dB (minus infinity dB) = absolute silence. Flat line in Audacity.
  3. Halving the linear amplitude = reducing by -6 dB. Doubling the linear amplitude = increasing by 6 dB

To explain the third one a bit:
“6” dB is a close approximation. The actual value is about -6.02059991328
Here’s some examples:

Linear Amplitude  | dB Amplitude
    1.0           |        0 dB
    0.5   (1/2)   |       -6 dB
    0.25  (1/4)   |       -12 dB
    0.125 (1/8)   |       -18 dB

Thanks Steve.

So I zoomed in on my waveform trying to guesstimate where threshold should be set, or where above I’d like to compress and I guessed 0.1 linear amplitude. Which is probably about -20 dB (yes, it is about 20 dBs after switching the scale on the track’s side scale read out to dBs). Though this seems a crude way to determine where to set the threshold… Is this the conventional way?

As well, the term compression is deceiving to me because I am always amplifying afterwards, raising the song back to where it was. So really (if I’m right) all compression, when volume is maxed afterwards, which is probably usually the case… is not so much “compression” as it is raising things from the threshold to the maximum peak. With “RMS” and “peak” compression doing the same thing only differing by the different parts of the waveform they focus on or raise to. So I’ll guess the threshold setting is not based on RMS but the peaks raised to (or compressed to) are RMS or peak depending on which one you choose.

There’s also mention in the manual of the compressor type chosen differences being that “peak” applies upward compression and “RMS” downward, but is this in reality just rhetoric though? Again, being that, if afterwards either is maxed to full volume then the only true difference would be the different parts of the waveform that were “compressed”, being RMS or dB?

Point being: Is it true, that there is no difference between upward or downward compression once makeup gain brings either to the others level?

So back to RMS compression and where to set the threshold… I did one edit using the SC4 maxed to “-30” threshold (using RMS at 1.4 ratio) and it did not sound good… It seems until I can learn a formula, or more of the knowledge as to where to set the threshold I’m best keeping with the formula I know: Highlight bulging spots on a track getting the RMS peak value, then take the highest and go a little above it… seems to work. Example, highest RMS peak on the track is -15.4 dBs, so I set the threshold to -15.0 decibels…

I’ll try to link the video where I learned this from if it’s still on YT. Yeah, here it is, what do you think of this guys summations Steve?
https://www.youtube.com/watch?v=lcEQJBO1Kj0

So, assuming the threshold sets at an amplitude value in the waveform as opposed to an RMS value, this formula I learnt may have some creedence… though too hard for me to wrap my head around at this point… Again, thanks Steve.

It’s far more conventional for compressors to work on peak level rather than RMS. Using RMS by default is one of the peculiarities of the Audacity compressor.

The word “compression” is referring to the “dynamic range”. That is, the amount of variation between the loud and quiet parts of the music.
A compressor reduces that range, either by making the loud parts quieter (sometimes called “downward” compression), or making the quiet parts louder (sometimes called “upward” compression), or both.

If all other factors are equal, then yes, if they are both brought to the same level then whether “upward” or “downward” the effect would be the same.

Personally I find “peak” compression much easier to use, and generally more effective than “RMS”.

Thanks Steve.

Wow, on my last edit (approach) there were peaks at 84 and 118 hertz I notch filtered 5 Q. Well first I noticed a lot of hum and I had already notched at 58 and previewed with (did not use) high pass which did nothing to stop the hum. Then luckily I checked the peaks around 100 hertz and the two mentioned definitely were the problem and they stood out a little from the rest of the 100 area.

Question: What could, humming around 100 hertz be coming from? I am in the USA and air conditioners are running everywhere, summer Phoenix, AZ. Of course I shut everything off in my place when recording but I share two walls with other units. Another possibility is my guitar humming, an acoustic, and or bouncing off the walls of the tight area I record in… It did not sound like guitar though, but did sound like air conditioning hum… Is it possible such hum can result in hitting the 100 hertz range?

“TASCAM DR5 PEAKY NOTES” UPDATE Aug. ?, 2016 - HOW TO DEAL WITH PEAKY NOTES by Black Dog Bluez

Recording my own acoustic guitar and vocal performances on one track, in one take, with one of the least expensive recorders, the Tascam DR-05. Then editing with Audacity. Results click here–> https://soundcloud.com/blackdogsongs

The Lo-Fi Challenge…

Recording with a Tascam DR5 stand-alone portable linear PCM recorder (Version 2 with current updates):

Well after all my Audacity trials and errors I now give up and am just releasing my songs straight with no editing… not sure what went wrong but it seemed the more I edited the worse things got, may be the recorder? May be the recorders mics? …I don’t know? It may even be Audacity? (??!) Anyway I now set my recorder’s settings to MP3 128k, what all the online players play at and will convert what ever you upload to. I also record in mono and set my mic input volume full up to “90” as high as it will go and use the built in limiter to keep it from clipping… This way I get the volume up enough without having to deal with the peaks (and editing)… I just watch to make sure I’m right around clipping, moving back and forth from the mic as necessary. This way everything is up, volume wise, and what goes too high gets reduced by the limiter.

It’s raw but pure. No compressor altering any of the nuances which is probably never good on intimate acoustic performances anyway. Though if volumes between lows and highs were a problem (or could be approached differently than limiting) I think this could be dealt with by something like an automatic fader of some sort… Which I think is a pay-only plugin exclusive to Pro-Tools (which I don’t have). Short of this, the limiter (including the Audacity Limiter) is probably the best choice from what I know …at this point.

Also It seems I have less floor noise compared to recording at a lower input level then having to edit everything back up (not sure on that though).

My recorder (the Tascam DR5) also has high pass filter which I don’t use because it sounds better without it. I do have some esses here and there though, which is about the only thing editing could improve … At this point, the less I have to tinker with a computer the better I like it, and probably the less the song gets degraded!

Then, I use MP3 Direct Cut (an MP3 non degrading editor) to trim and fade the ends, and check the volumes. Sometimes lowering a peak or two …or the overall volume when there are excessive peaks hitting zero and ‘normalizing’ shows song to be maxed at zero. I also confirm/check volume (number/value) with MP3 Gain (93 or higher is sufficient volume IMO). Unfortunately MP3directcut does not give a volume value.

NOTE: Sometimes MP3gain will show a song to be clipping when MP3directcut does not. MP3diectcut will show peaks to be below zero - ‘not clipping’ with it’s “normalize” feature - which allows you to evaluate and/or adjust volumes. This volume discrepency I assume is just a difference in exactly where these programs consider clipping to be … I side with MP3directcut’s analysis as the final authority on what is clipping.

Will update if things change.

FYI: Check your MP3s with the free downloadable freeware “MP3gain” and for confirmation of “clipping” — “MP3directcut”… you’ll be suprised how much ear numbing distortion you’ve been listening to…

Section 3E5.2:

Using multiples of six as the value of each letter of the alphabet… A = 6, B = 12, etc…

C 3 O 15 M 13 P 16 U 21 T 20 E 5 R 18 = 111 x 6 = 666!

In closing, demand grid-less electro-smog-less life! Getting old does not suck, getting stupid does. I’m fifty and I can still touch my knees! … If you’ve read this far you have the attention span of a salamander … I like that!; Thanks… to all the NSA computers ensuring none of us are tourists… a real world-wide threat. Flying everywhere, never happy to be here now… So that is the challenge folks! To be here now, as you are. Thanks for that… being here now… although by now I’m not here with you… I’ve become a bird and have flown the coupe …that’s me outside your window now.

We become unhealthy-techno junkies the further we are removed from nature.

Black Dog Bluez… (Lost Here … and going mad!) https://soundcloud.com/blackdogsongs