Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Confusion with normalizing (Read 5459 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Confusion with normalizing

I have a huge music collection, and I prefer listening to CD's. When I'm not at home, I listen to mp3 rips on my phone normalized with MP3Gain. But I'm quite confused about normalizing.

I know most modern recordings found on CD's are mastered very loud and that they clip frequently. But the reason I was normalizing my mp3's was to don't have to adjust the volume when playing playlists containing tracks with big differences in volume. I was not thinking about this whole clipping thing.

Let's say I play a CD which is mastered very loud, and then I play a flac rip of the CD which is normalized so that it's not clipping (Whatever that means). I find two different volumes on my amp for both the CD and the flac files which makes them equally as loud. What will sound best? Will the flac files sound better because they are normalized so that they don't clip? I thought they would sound the same since I matched the volume, but reading all this about normalizing and clipping got me wondering. Shouldn't the CD and the rips sound the same because the dynamic range is still the same (obviously)? I'm reading alot about people having mp3s that's clipping, and that they're "fixing" them by normalizing them so that they're not clipping. What do you actually "fix" by normalizing? The dynamics from the recording is crushed by mastering anyways, right? 

Could someone educated please enlighten me about this? I have actually searched alot online, but can't find an answer to this specific question.

Thanks.

Confusion with normalizing

Reply #1
Let's say I play a CD which is mastered very loud, and then I play a flac rip of the CD which is normalized so that it's not clipping (Whatever that means).


If the CD is clipping, then so will the FLAC. 

I'm reading alot about people having mp3s that's clipping, and that they're "fixing" them by normalizing them so that they're not clipping. What do you actually "fix" by normalizing? The dynamics from the recording is crushed by mastering anyways, right?


MP3 is a lossy codec, so if the file is very loud when you compress it, it may clip on decoding, although this is rarely audible.  This is not really related to a CD clipping directly, although a CD that is clipping is more likely to have additional clipping if encoded to lossy format.


Confusion with normalizing

Reply #2
Let's say I play a CD which is mastered very loud, and then I play a flac rip of the CD which is normalized so that it's not clipping (Whatever that means).


If the CD is clipping, then so will the FLAC. 

I'm reading alot about people having mp3s that's clipping, and that they're "fixing" them by normalizing them so that they're not clipping. What do you actually "fix" by normalizing? The dynamics from the recording is crushed by mastering anyways, right?


MP3 is a lossy codec, so if the file is very loud when you compress it, it may clip on decoding, although this is rarely audible.  This is not really related to a CD clipping directly, although a CD that is clipping is more likely to have additional clipping if encoded to lossy format.

So the clipping we're talking about when normalizing an mp3 is this additional clipping made by compression, and not the clipping from the result of loud mastering?

Confusion with normalizing

Reply #3
So the clipping we're talking about when normalizing an mp3 is this additional clipping made by compression

No.  The encoding process is generally not the cause of additional clipping. The additional clipping happens when decoding data to integer PCM that exceeds 0dBFS.  Lowering the gain of the encoded data so that the peaks are below 0dBFS or decoding to float will prevent additional clipping.

clipping from the result of loud mastering

It's generally accepted that loud mastering exacerbates the problem.  The fortunate part is that replaygain will usually provide more than enough headroom for these peaks because of the drop in level.

To support saratoga's point, I know of no evidence to support the fear that the audibility of this additional clipping is pervasive with real-world music.  There's a lot of useless anecdotal talk on the subject, however.

Confusion with normalizing

Reply #4
So the clipping we're talking about when normalizing an mp3 is this additional clipping made by compression

No.  The encoding process is generally not the cause of additional clipping. The additional clipping happens when decoding data to integer PCM that exceeds 0dBFS.  Lowering the gain of the encoded data so that the peaks are below 0dBFS or decoding to float will prevent additional clipping.

clipping from the result of loud mastering

It's generally accepted that loud mastering exacerbates the problem.  The fortunate part is that replaygain will usually provide more than enough headroom for these peaks because of the drop in level.

To support saratoga's point, I know of no evidence to support the fear that the audibility of this additional clipping is pervasive with real-world music.  There's a lot of useless anecdotal talk on the subject, however.

Ok, that cleared things up! If I'm understanding everything correctly, the additional clipping is in lossy formats only like mp3 and aac? I don't need to normalize flacs from CD's to prevent clipping, right?

Confusion with normalizing

Reply #5
The additional clipping is not "in" the lossy formats, it only happens when decoding these lossy formats to integer PCM if the decoded data goes beyond 0dBFS.

If there was additional clipping in lossless formats they wouldn't be lossless.

Instead of worrying over inaudible clipping upon decoding, I would worry about whether I have any highly dynamic content (lossy or lossless) that would be driven into audible clipping by positive gain adjustments.

Confusion with normalizing

Reply #6
The additional clipping is not "in" the lossy formats, it only happens when decoding these lossy formats to integer PCM if the decoded data goes beyond 0dBFS.

If there was additional clipping in lossless formats they wouldn't be lossless.

Instead of worrying over inaudible clipping upon decoding, I would worry about whether I have any highly dynamic content (lossy or lossless) that would be driven into audible clipping by positive gain adjustments.

I do have some old CD's that are very quiet and has good dynamics compared to modern ones. Would mp3 rips from a CD like that need positive gain from MP3Gain? (If that's what you meant by your last sentence) Will clipping be more audible in mp3's with high dynamics? Is 89 dB in MP3Gain fine?

Confusion with normalizing

Reply #7
"need positive gain"
You mean to say result in positive gain.  They may, yes.  Lowering the reference to the original value of 83 should probably prevent this, though MP3Gain will tell you whether they clip (at least the GUI version will).

I don't wish to speculate about audibility.  Speculating that dynamically compressed lossy-encoded music is more likely to result in clipping upon decoding to integer should go far enough to answer that portion of the clipping question, if that's the angle you're taking.

Confusion with normalizing

Reply #8
Thanks for great answers, guys!

Confusion with normalizing

Reply #9
I don't need to normalize flacs from CD's to prevent clipping, right?
If the CD has clipping, the FLAC will have clipping, and there's nothing you can do about it. Reducing the level simply makes the flat (clipped) waveform tops play back at a slightly lower amplitude (i.e. not at digital full scale), but they're still flat (clipped). You haven't improved anything.

Cheers,
David.