Tips for a live recording: how to Normalize properly

Hi,
Im relatively new to audio processing. I’ve been using Audacity for years but not for effects and manipulating the sound, so im looking for any advice from this group about anything useful and relevant to processing live concert recordings. I have captured a bunch and would love to get your advice on how to improve them by using some of Audacity’s features, such as Normalize, Amplify, EQ, etc.

I have searched and found so many random posts and articles but each one is pertaining to its own context and hard to extract anything useful for my situation or with enough information to try things correctly with the correct settings. So thanks in advance.

For example, i’ve been told that i should Normalize my entire recording once i’ve finished trimming and editing it. This effect tries to bring the highest and lowest levels closer together creating a more consistent or “normal” level of audio. Is that right? There are settings in this feature and I have no idea what numbers to put in there. someone said to find my lowest level and use that as a reference, but i dont know how.

Everyone has different techniques and approaches, so it would great to hear them, and decide what works best for me.

Thanks.

No, that’s not correct.
“Normalize” just amplifies the audio to a specified level.

[Quick definition: “0 dB” represents “full scale” - the maximum signal level. Signal levels are thus measured as “negative dB”, which is an indication of how much below “full scale” the signal is. Absolute silence is “negative infinity dB”. ]

You would usually want your recordings to have a peak level close to 0 dB (full track height). Commercial CDs are usually normalized so that the peak level is exactly 0 dB, though for best quality it is usually advised to allow a little “head room” as hardly any audio systems can handle signals up to 0 dB without significantly increased distortion. For this reason, it is often advised to normalize to around -1 dB for lossless formats (such as WAV), or a little lower for lossy formats (such as MP3).

Audacity’s “Normalize” effect defaults to -1 dB.


To “bring the highest and lowest levels closer together” is called “dynamic range compression” (often called simply “compression”, though it should not be confused with “data compression” which is about reducing the size of files). The two types of effect that produce “dynamic range compression” are “Limiters” and “Compressors”.

“Dynamic range” is (broadly speaking) the range between loud sounds and quiet sounds.

Compressors” act over a large portion of this range. They may work by raising the level of quiet sounds (“upward compression”), or reducing the level of loud sounds (“downward compression”), or a combination of the two. Compressors generally adjust the level relatively slowly, and can be visualised as adjusting the volume control of your amp while the music is playing (turning the volume down when the sound gets loud, and turning it up when the sound becomes quiet.

Limiters” generally act only on the highest peaks of the waveform, leaving the rest of the audio unaffected. In effect, they “squash” the peaks, so reducing the difference between the peaks and the level of the rest of the audio (reducing the dynamic range).

Many natural sounds (including acoustic instruments) can have very high peaks compared to the overall level of the sound. When such a recording is normalized, the recording may sound rather quiet, because the peaks are so much higher than the rest of the audio. This is a common use for Limiters. The Limiter squashes down the peaks to a more manageable level, which allows the recording to be amplified a bit more without the peaks exceeding 0 dB.


A fairly standard workflow when recording live music as a single track:

  1. Record, aiming for a maximum peak level of around half the track height. In digital audio, the exact level is unimportant but it MUST remain below 0 dB at all times (otherwise there will be irreparable “clipping” distortion).
  2. Normalize to around -1 dB. This gives a nice big waveform to work with.
  3. Do all of your editing and processing.
  4. Normalize to 0 dB.
    Although the peaks are at max, the overall volume is probably a little low.
  5. Apply a Limiter effect, to squash the peaks down a bit (such as Audacity’s limiter: Limiter - Audacity Manual)
  6. Normalize to -1 dB, and export.

If working with multiple tracks, you will need to mix down the tracks (“Tracks menu > Mix > Mix and Render”) before the final “Limiter / Normalize” steps.

this is SO HELPFUL. exactly the kind of fast definitions i was looking for.

Is there anything i need to know about the Limiter settings in my application? to bring down some of the peaks, like clappng/applause, whoops, etc? Those are the main peaks, otherwise the music itself is pretty even range (until i make my waveform bigger and see the actual range.)

the only thing i know about limiting so far is on my recording software (shure MV88 mic and app) i ruined very nice recording of a show because the device/app had defaulting to having the limiter ON and so my recording was squashed down everytime the kick drum hit or the bass guitar thumped, so pretty much the whole show. The result was a recording where the dynamic range was effected and the limiter took away most of the bass and range every beat of the songs due to the kick drum. And i know theres settings to that too… The attack and the dissolve of that effect right?

Thanks again.
jeff

Personally, I like to avoid using too much dynamic compression. Modern commercial recordings are typically compressed to extremes, which in my opinion, robs them of much of the life and excitement of live performances.

The default settings in Audacity’s Limiter are generally a good starting point. Effective for dealing with occasional peaks, but not too severe.

For reducing the level of applause between songs, the “Envelope Tool” is probably best: https://manual.audacityteam.org/man/envelope_tool.html

Thanks tons for this, Steve. I am playing around now, and actually feel like i know what I’m doing.

I do have experience with the envelope tool. That is helpful in some of my quiet room performances that have loud songs and talking in between them.
thanks.
jeff

The hardest part is getting a good recording to start with. Typically, there is too much room reverb and room noise. The amount of reverb that sounds great in a concert hall usually sounds unnatural in a living room, and the background noise that you may not notice live tends to stand-out in a recording. (Although after you have experience recording you tend to become more-aware of the noises.) Professional live rock recordings are close-mic’d and multitracked (with separate mics/tracks for the audience) so they are almost getting a studio recording.

Is there anything i need to know about the Limiter settings in my application? to bring down some of the peaks, like clappng/applause, whoops, etc?

Limiting probably won’t make those parts sound much quieter.

So yes, it’s probably the best thing to start with the Envelope Tool. The trick is to fade-down and fade-up so you’re not making any sudden-unnatural changes in volume. Then you can normalize to (possibly) bring-up the overall loudness.

If it’s not loud enough overall, you can use the limiter (with make-up gain to bring-up the loudness). Before limiting I recommend normalizing (or just use the Amplify effect at the default setting) to get a known “starting point”. (Normalizing as the last step after everything is usually a good idea too.)

The Audacity limiter is very good. It uses look-ahead so the hard-limit setting won’t distort the waveform. If the original recording isn’t clipped you can probably “get away” with 6dB of limiting with very little effect on the sound quality. You can repeat the process if you want more of the effect.

Note that compression and limiting make the signal-to-noise ratio worse and since they are normally used with make-up gain, the background noise increases. That’s just something you have to live with except where you want to turn everything down (between songs, etc.) (It’s OK to use make-up gain… It makes no difference if you add the gain or if the listener turns-up the volume control.)

Limiting is a kind of (fast) dynamic compression. You can also use “regular” compression, but there are more settings to mess with (and different compressor plug-ins that you can install) so things are easier if you can get good results with just limiting.


It’s up to you if you want to process the songs individually or the performance as a whole. If you process the songs individually, re-join with a crossfade so the “splice” is not noticeable.

Even if you process the concert as a whole, you may wan to edit-out excessive gaps or extra talking between songs. By crossfading the crowd noise you can keep the recording natural sounding without long gaps between songs.

Often, I’ll “steal” applause from one part of the performance and mix it back-in different places. Depending on what I’m working with I like to “build” one or more applause tracks of 20 seconds or longer, then fade/edit/mix with the original recording where needed. A couple of times I’ve stolen applause from a different recording!

If I’m making separate files for each live song, I’ll also use some applause “tricks”. I general like a short 1-2 second applause/crowd noise fade-in at the beginning of the song and 10-15 seconds of applause at the end with the last half (or more) fading-out.

But just a note… I’m rarely working with original recordings… I’ve done this when making CDs or MP3 from professionally recorded video concerts.