What is the difference between the Normalize effect and the Loudness normalizer effect? In layman’s terms. I would explain it as: the Normalize effect is about peaks, and the Loudness normalizer effect is about averages. Am I close?
Thx in advance!
PS. If I have a track, with me talking, how can I measure that audio track’s RMS and peak?
I cannot find the EDIT-button. Just want to add that I have the latest version (2.4.2).
loudness normalizers/equalizers take into account that the human ear is not equally sensitive to all frequencies
It just feels like you’re trying to read for audiobooks. Did I hit it?
Effect > Amplify works from blue wave tips and peaks. Effect > Normalize used to as well with a couple of additional variations thrown in such as DC removal. Loudness Normalize is very different. The two variations are RMS and LUFS.
RMS is pretty simple. It’s an electrical measurement of horsepower energy in a wav. It’s always lower value than peak and it’s the lighter blue pattern in the Audacity timeline. Electric motors don’t care at all what the peak values of their power are. They need the RMS energy to turn.
You can measure your performance RMS value with Analyze > Contrast > Measure Selection.
It’s also, usually much more importantly, how ACX measures your audiobook voice volume.
LUFS or Loudness Unit Full Scale is similar to RMS, except it takes into account your ear sensitivity. That one is the difference between low rolling thunder on a summer evening and baby screaming on a jet. The thunder is louder, but LUFS isn’t fooled for a minute.
Audacity publishes ACX Check which is a tool that measures the three Audiobook technical specifications. Peak, RMS and Noise.
We publish other audiobook tools, too.
“Normalize” is about peaks (you were spot on with that one )
“Loudness Normalization” is about “perceived loudness”. The idea here is that if you normalize lots of music tracks to the same “loudness” (measured in LUFS), then they will all sound (approximately) as loud as each other. Radio stations typically do this so that they don’t have sudden jumps in volume from one track to the next. Note that when normalizing loudness, it is important to allow a lot of headroom on “loud” (heavily compressed) tracks, otherwise it will not be possible to make tracks that have greater dynamic range sound as loud without clipping (distorting) the peaks.
See also: https://manual.audacityteam.org/man/loudness_normalization.html
Normalize effect is about peaks, and the Loudness normalizer effect is about averages. Am I close?
Yes, regular normalization is based on the peaks (usually setting the peaks to 0dB (= 100% = 1.0). The peaks don’t correlate well with perceived loudness. If you normalize all of your music some songs will still be louder than others. (Most commercial music is already normalized.)
[u]RMS[/u] is a kind of average and it correlates better with perceived loudness. The ACX audiobook standards use RMS.
If you calculate the simple average of an audio file it’s zero because the waveform is positive half the time and negative half the time.
[u]LUFS/EBU R128[/u] does even better than RMS by taking the frequency content into account. (i.e. a 0dB 2kHz tone sounds louder than a 0dB 50Hz or 10kHz tone.)
…You have to be careful with RMS or LUFS normalization because you an push the peaks into clipping.
Many thanks for that educational link. It had a lot of new and relevant info. about loudness.
On a side-note, I’ve always wondered what Hz humans are most sensitive for — that text said between 2,000 and 5,000 Hz so now I know
Yes, it’s about audiobooks, just wanted to learn more. Not narrating anything atm. I had never noticed that the waveform had a core. Now I know it’s rms. Many thanks for the info and links about ACX Check, will take a closer look tomorrow. You mention other tools, are they super important for beginners, or can I learn those later on? All the best!
You mention other tools, are they super important for beginners, or can I learn those later on?
There is a [u]Recommended Audiobook Mastering Process[/u]. You RMS normalize and then run the limiter to get the peaks in-spec and that takes care of your RMS & peak levels every time.
Noise is usually a bigger issue since most people don’t have a soundproof recording studio in their home. (The equipment has to be reasonably-good quality too.) RMS normalization usually boosts the volume so it will also boost the noise.
You mention other tools, are they super important for beginners
There are no audiobook beginners. ACX, and I understand the other publishers, expect you to be “Retail Ready” when you send them voice work. ACX publishes help pages and videos, but isn’t interested in teaching you one-on-one how to read.
It’s something of shock the first time you find out what you have to go through to get a quiet, echo-free recording space. The microphone makers want you to think you can set up their microphone on the kitchen table and read for publication.
Not, usually, no. Some people read successful books using actual paid professional soundstudios. They appear on the forum when they try to do it at home and find it’s not as easy as it looks.
Let us know.
Noise is usually a bigger issue since most people don’t have a soundproof recording studio
That’s the obvious one. You may be able to get around that with tricks. Many microphones come out of the box making noises of their own and some of those are pitched down in the earthquake/thunder range. That’s why the first step in Audiobook Mastering is “Low Rolloff.” It removes low pitch tones that you can’t hear, but mess up the other settings. Some of those tones on some microphones can be louder than your voice.
That’s also why you want to apply the mastering tools in order, don’t add any tools, and don’t leave any out.
You can submit a short voice test to the forum…
…and you can submit a different one to ACX, too, but it may take them weeks to get back to you. They’re running in overload and they said so in their posted instructions.
You can use Audacity ACX Check to test for technical compliance. It’s a similar test to The Robot that ACX uses when your submission first arrives, but then ACX goes on to Human Quality Control where you get judged on voice quality and theater excellence. Human QC is where you go to die if you have voice defects, stutter, P-Popping, or spitting on the microphone. This is also where you die if you had to heavily process your voice to make it past technical compliance. Noise Reduction, Noise Gate, Compressors, Silence Editing, etc.
Nobody is going to pay to hear you read a story in cellphone voice.
You can read to your own podcast, roundtable or conference however you want, but you have to pass a lot of serious testing to get your audobook published.
And that’s just the sound part. ACX is also sticky about you posting rights and permissions to the book. There is no reading somebody’s book without telling them you’re doing it. Even if it’s your book, you have to prove you didn’t promise rights to anybody else.
There is one other “fuzzy” oddity. The hiking consideration. I like putting in my earbuds climbing into my backpack and hiking around the wilds of Los Angeles listening to Sarah Vowell read me a story. We had two posters on the forum who didn’t fit that description. One was reading a cookbook and the other was reading meditation lessons. I couldn’t picture myself crossing Wilshire Blvd and “releasing my tensions” at the same time. Neither one ever posted back how it went, but we have had people post their successes. So I’m assuming it didn’t work out.
ACX hates music and they don’t do radio theater.
So no, there are no beginners.