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: Difference between sound played on YouTube+same song extracted to MP4? (Read 16401 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #25 –
Hello again,

Generally it looks like you are right - the sound streamed on-line via YT is "simply" louder than this played from hard drive. Now I would like to better understand this phenomen from technical point of view.

It's an oscillogram graph of track extracted from YT file (view in GoldWave):



I don't know what is the scale measure of OY axis, but I see that the sound doesn't fill the full range of the graph. It oscillates from about -0,4 to 0,4 instead of -1.0 to 1.0.

Now I did an experiment and maximized the volume with GoldWave to so called "half dynamic range". The new graph is below:



Now, the file streamed from YT, and the same played with player sounds almost identically.
I understand, that YT has to process the played sound on the fly.

You've asked me about my AIMP3 player options. To this day I used options as on two next screens:





Today, I've found very interesting option in DSP Manager. It's connected with volume normalization and it was called "Normalize by amplitude". Checking this option causes radical increase of volume (sometimes more than track played on YT).

I'd like to take a knowledge about these two things especially:
1) What kind of volume normalization does YT use?
2) I'd like to record YT tracks to CD-Audio with the same volume level as in YT. What I need to do?

I've googled for some info to solve my problems, but I'm not sure if I understand correct.

We can use two volume normalization algorithms at least:
A) Peak Normalization
B) GainReplay Normalization

a.d A) I understand that Peak Normalization simply multiplies each sample by some factor to increase the volume to desired volume level. Do I understand correctly, that it's the equivalent of matching/maximizing sound volume in GoldWave?
a.d B) I understand that GainReplay has two modes: track and album. Track mode tends to adjust a track volume level to average track volume level (so each sample is multiplied by different factor) and the album mode tends to adjust a track volume level to average album volume level. Is it correct?

I suspect that this issue is connected with so called "dynamic range compression" (I know that YT decreases the track dynamic range, what meets with users and artists protest), but currently I have too small knowledge in this topic, to understand it.

According to my question number two I've tried to normalize ripped track volume level in Easy CD-DA Extractor, but I don't understand. Please see the attached image:



Why, when I put number 100 (I think: 100%), the volume level will be decreased (-11 dB)? I'm sorry if this question sounds stupid for you, but I really don't understand this matter - maybe due to lack in my knowledge at this topic.

Thank you for your help and forbearance.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #26 –
I wouldn’t be so quick to rush to the conclusion that YouTube applies specific normalisation and start trying to derive formulae, etc. It could just as well be the case that it simply has a volume control that goes a bit past 0 dB/full scale when at maximum.

FWIW, I quite often get clipping while watching videos on YouTube and have to turn its volume down, although previously I was more likely to attribute this to the abysmal computer I currently use, and I would still not like to make any confident pronouncements either way; still, here it is for the record.

Edit:
Quote from: Ulisses on
http://www.ists.pl/~michmajsoftware/Private/pic5.png

Why, when I put number 100 (I think: 100%), the volume level will be decreased (-11 dB)? I'm sorry if this question sounds stupid for you, but I really don't understand this matter - maybe due to lack in my knowledge at this topic.
It seems fairly clear to me that the number in that box is to be read as dB, not a percentage. In that context, 100 dB refers to SPL, not digital gain/attenuation, and that differs from the usual 89 dB target volume of ReplayGain by 11 dB. That explains both figures.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #27 –
Quote
simply has a volume control that goes a bit past 0 dB/full scale when at maximum

Not informing the user at what location 0dbFS is....is dumb.
I have noticed clipping on lots of Youtube clips before, but never thought to check the volume. I thought it was just bad uploads

Never understood why VLC media player could go past 100%. Everytime I went past 100% I got horrible clipping/distortions.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #28 –
Quote from: LithosZA on
Quote
simply has a volume control that goes a bit past 0 dB/full scale when at maximum

Not informing the user at what location 0dbFS is....is dumb.
I have noticed clipping on lots of Youtube clips before, but never thought to check the volume. I thought it was just bad uploads
Just to be clear, the section of my post that is quoted here is prefaced by the condition that this is not something I know or have evidence for, just a possibility. I’m not going to claim (as anything more than a mere possibility) that YouTube definitely allows amplification past 0 dB when I can’t confidently disentangle my experiences of clipping on YouTube from my questionable onboard audio hardware or Windows’ subsystem overall; moreover, I tend to do enough other audio-based things on this PC too rarely to say whether or not other applications can exhibit similar effects.

Quote
Never understood why VLC media player could go past 100%. Everytime I went past 100% I got horrible clipping/distortions.
It depends upon both the source material and your hardware. I questioned the feature at first but have come to find it hugely useful for certain things like quietly mastered recordings of concerts, etc. where 100% is nowhere near enough and adding volume would otherwise be quite laborious and probably involve some other application interfering with things (guesswork as I’ve never needed to search for a method).

Difference between sound played on YouTube+same song extracted to MP4?

Reply #29 –
Yes, sometimes dialogue levels for films are so quiet that I want to make it louder.
Who cares clipping of BANG!!!! BOOOOM!!! when dialogues are hard to hear?
Probably some kind of dynamic range compression is more desirable.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #30 –
Quote
I wouldn’t be so quick to rush to the conclusion that YouTube applies specific normalisation and start trying to derive formulae, etc. It could just as well be the case that it simply has a volume control that goes a bit past 0 dB/full scale when at maximum.


It was my first think, bu I like to complicate my life, so I'm trying to create more sophisticated theory

So, we have two hypothesis:

1) YouTube uses volume slider over 0 dbFS, what can be heard as so called clipping, especially when a track peak has 100% of scale at given moment. It's very interesting kind of volume normalization in such case...

2) YouTube uses some kind of dynamic range compression (maybe in real time, during streaming). BTW: I 2008 YT committed that YT was using dynamic range compression (source1), but seven days later, after many protests, YT decided to stop this proceeder (source2)

Unfortunately I think that it's impossible to discover the truth with 100% sureness, without possibility of looking inside YT player mechanism.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #31 –
Quote from: Ulisses on
2) YouTube uses some kind of dynamic range compression (maybe in real time, during streaming). BTW: I 2008 YT committed that YT was using dynamic range compression (source1), but seven days later, after many protests, YT decided to stop this proceeder (source2)


Well, first link is malformed, let's correct it for  convenience :source1
According to your source, we have the proof that youtube applied at least once loudness compression.
Now who know if they completely renounced to perform loudness compression, or  if they apply it on only particular cases.

Difference between sound played on YouTube+same song extracted to MP4?

Reply #32 –
dynamic compression during uploading/encoding doesn't matter if we compare the sound from web browser vs the sound from downloaded mp4 file.