Hard Limiting VS Normalizing

episode

New member
Hard Limiting & Normalizing

what exactly is the difference? i am a vinyl dj who plans to transfer all of my music to CDs. using cool edit pro, i want to record at -6 dB and boost the amplitude up. should i use Hard Limiting VS Normalizing? what does each one do and what will be better to retain all of the originalaudio qualities of the song? thanx.

episode
 
The normal use of normalise is peak normalising. What it does is look at the signal, find the loudest peak, and amplify/decrease the signal till the loudest peak will be at the value you set.

For example, if the loudest peak of your recorded signal is -6 dB, and you asked to normalise to 0 dB, it will amplify the whole signal 6 dB. So normalising is basically just an amplification.
Make it a habit to normalise to -0.1 or -0.2 dB, as some cdplayers see 0 dB as errors.

Some programs have also a features called RMS normalising (like Soundforge). This one acts on the RMS value (kind of the average volume of the signal). It does the same, looks at RMS value, and amplifies until it reaches the value you set. Frankly, I wouldn't use RMS normalising, as it doesn't prevent clipping, and seriously, compression is more versatile here.

Hard limiting is a steep form of compression. A compression with a very high ratio (if it's very very high, we also speak of brickwall limiting).
When you set a high ratio, it means that even if a high signal comes at the input, not much extra signal is presented at the output. Generally, a very fast attack and release is set, so it only acts on the peaks (and transients). Brickwall limiting is the extreme here, as you set a certain treshold, and nothing that goes over this threshold can pass.

So with hard limiting, when it's used as loudness maximizer, you generally set a maximum output level. For the sake of ease, we'll take -0.2 dB like with normalising, and I'll take brickwall limiting as an example here. Now you can amplify the signal (either with the make up gain of the compressor/limiter, or with amplification pre compressor), but the maximum output will not pass -0.2 dB. Your overall level will increase, but you won't clip.
If you use a ratio a little less than brickwall, it means that some very high peaks at the input, will still pass -0.2 dB at the output.

So, when you read and understand the working of both methods, you'll see that normalising will be the less harming to the original signal of the two. You just amplify the signal so it maximizes at the wanted value. You don't exactly change anything to the waveform (except amplitude).

With limiting, you change the waveforms. If you look at them in an audio editor, after you limited, you'll see that the peaks of the signal won't be smooth anymore, but they'll look like they will have been cut off (and sadly this is a practice much done in the commercial music circuit nowadays). Limiting should only be done very mildly, for the uttermost peaks. Don't try to gain 6 dB with a limiter. Maybe only the best like Weiss will still sound natural, but for all the rest... It might sound louder overall, but also more tiring to the ears.

So if you want to record at -6 dB, fine. Leaves you margin enough not to clip. After you've done editing (if any), normalise close to 0 dB (normalizing should be done as one of the last steps, prior to dithering if you reduce bitdepth).

If after normalisation you find it's not loud enough overall (shame on you :D), I would recommend another method that is more easy on the ears, and used in some mastering houses to increase loudness : use a compressor with a very low ratio. Like 1.01:1, I'd say maximum 1.2:1. Adjust the threshold until you got the gain reduction you want (trial and error, and practice will tell you how much). Like if you want 3 dB increase of loudness, adjust the treshold until you got about 3 dB gain reduction. Usually short attack, longer release. Listen for pumping of course. Then adjust make up gain (if it's not automatic like on some plugins) until you get the right level again.

This is a method that is preferred to hard limiting. The very low threshold ensures the compressor is working on most of the signal, and the very low ratio ensures you don't actually hear very much compression, and doesn't alter the waveforms like limiting does, in an unnatural way. It will be more pleasing to the ear, and you still got your increase in loudness.
 
Good advice Thy.
Just to add, try to record as 'hot' a signal as you can as normalising also raises the whole signal path by the selected amount which means you also raise the noise floor level and lower the headroom.
 
Back
Top