Novice question: should I normalize my audiobook sound?

Question

I have ripped, using Windows Media Player, the audio from an audiobook CD. Now I am using Audacity to make some slight clipping changes to the tracks.

The Audacity manual recommends -- for example, here and here -- that audio be normalized (especially when ripping from an LP or cassette tape, which I'm not doing here). Should I normalize the audio that I ripped from an audiobook CD?

I am a sound editing novice, so -- not knowing what to do -- I tried normalizing my track using the default settings in Audacity:

Here is the waveform of the track before normalization:

BEFORE

... and here it is after normalization:

AFTER

Which of these -- BEFORE or AFTER -- is "better"?

Clearly, this is a vague question, since what is "better" probably depends on the application. But since I'm a novice, please bear with me and try to give me some advice.

From what I understand, normalization is the process of scaling the maximum wave amplitude to a specified decibel level. This, I think, somewhat alters the "loudness" of a recording. I think that this can be useful if multiple tracks were originally recorded at different levels; you generally don't want a really "loud" song followed by a really "soft" song on the next track.

But if all of the tracks ripped from the audiobook CD appear to already have the same maximum amplitude (i.e., all tracks of the audiobook were recorded at the same level), is there any reason to normalize all of the tracks to a new maximum amplitude, such as the -1.0 dB that Audacity recommends as a default setting?

And suppose I later rip another, different audiobook CD that was recorded at a different level; would it be a good idea to normalize that audiobook to the same -1.0 dB level as the first one?

I will be playing the audiobook(s) on a smartphone or a tablet, just for personal use.

Answer 1

Normalization is a very simple process. It finds the highest peak and applies gain (makes everything louder) until the highest peak is at the specified maximum level (-1dB in the picture).

There's also the option to remove DC. You can imagine DC as an upward or downward shift in the waveform, making it stay averagely more on the positive or the negative side. This has the effect that when the software looks for the highest peak and amplifies, the other (lower in volume) end of the waveform will not reach the maximum level, leaving unused dynamic range.

So it is always a good idea to remove DC when normalizing.

So, to answer your question directly, better means louder and you can see this in your pictures. If you always normalize your audiobooks, then all of them will sound approximately at the same level.

I'm saying 'approximately' because the perceived level is different than the actual level.

Answer 2

Normalization of a digital audio source adds renormalization noise. If you are still working with 24bit from a digital master, the effects are negligible. For 16bit->16bit or less, not so much. Such noise can be rendered less conspicuous (but taking up more sound energy) using dithering.

For professionally mastered CD I copy to CD I would refrain from renormalization. If the source and end format differ anyway (different sampling rate, MP3 or other compression), it's a tossup and you might as well normalize. However, it still makes sense to properly "gain-stage" it: this usually means that if you have a format/rate-conversion process involved that you convert to a higher-bitdepth target format in the process and then normalize only as part of the final reduction step.

Answer 3

You should adjust your sound to match the LUFs or equivalent specification of where your audiobook sound will be used to deliver it.

LUFs is a loudness standard that includes peak and average energy with its duration so as to avoid the loudness wars or problems as often is heard with commercials or self made RockType music CDs.

Normalization for LUFs is different than audacity normalization which merely raises the signal so the peak point is at (default) -1dB