Fre:ac vs. Audacity MP3 Encoding and NoiseFloor Values ?

[SheilaQ]

Good afternoon all,

OK, here's my newest question -- in an attempt to try and better understand some of the more technical aspects -- though I have a LOOOOOONG way to go and much of which I'll leave to you guys! Honestly, just looking to be comfortable and knowledgeable about what I need to know for audiobook and ACX-related elements within Audacity.

So, I took a .aup file (ran the ACX Check on it) and then performed the MP3 encoding using fre:ac (with the instructions/settings I found on the ACX blog - which still uses LAME), I then created an MP3 directly from Audacity (LAME). Afterwards, I ran the ACX Check on the two MP3 files to see how they all differed. Attached are the screen shots of the results for each. It clearly shows how the MP3 encoding can raise the peaks as I normalized the original .aup file at -3.2dB.

Also, I'd appreciate it if someone would explain to me what the two separate NoiseFloor results are (in the ACX Check results) and why they are seemingly quite far apart? So, that's it for this post. I always appreciate you guys' time and expertise and for always being so willing to help all of us and explain things. Thanks a ton!

Oh, just to confirm for this post's sake:

Windows 7 Pro -- Audacity 2.1.2
AT-2035 Mic
Focusrite Scarlett 2i2 audio interface (1st gen)

Sheila

[DVDdoug]

It clearly shows how the MP3 encoding can raise the peaks as I normalized the original .aup file at -3.2dB.

I see your original level of -3.2dB and an MP3 peak of -3.5. These are negative numbers so with this particular file MP3 encoding REDUCED the peaks by 0.3dB. (The RMS level is also slightly reduced.)

Also, I'd appreciate it if someone would explain to me what the two separate NoiseFloor results are (in the ACX Check results) and why they are seemingly quite far apart?

A Weighting takes human hearing into account. Your ear is less sensitive at very-high and very-low frequencies. Any noise at the extremes won't be as noticeable (some noise might even be completely beyond the human hearing range).

Since these frequency extremes are not as easily heard, they are reduced in the A-weighted measurement and the A-weighted measurement gives you a "better number" and it's a better (more useful) measurement.

[SheilaQ]

Ahhhhh! Thanks so much for the explanation on the NoiseFloor levels. And, yes, thanks too for correcting me on the peak numbers. I still have difficulty getting it right in my brain about how the negative number thing works. That's another item I'll try to keep front an center in my brain to try and get better with. Thanks so much again for your explanation. Hope you're having a great weekend!! Take care . . .

Sheila