Normalize or Replay-Gain ?

[Richx200]

Hello,

I consolidated numerous tracks from different sources into various configurations. As a result, I find myself holding the remote control to change the volume for each track.

I have a number of options to level the tracks, I just don't know which leveling would work the best.

First, there is normalizing the tracks, so the loudest peak is equal to a percentage of the track's maximum signal.
With this effect, I can use Peak Normalization, Average Loudness, Peak Loudness, Perceived Loudness (DBA) or Integrated Loudness (EBU in The R-128 standard)

The other method would be Replay Gain (my streaming device does support Replay Gain) The program PerfectTunes; "By Fault Track and Album gain is written into the track's Tab calculated with EBR128 and a LUFS of -18 target volume"

Normalization will change +/- the music's, volume while Replay Gain will put a +/- db and peak db in the tracks tab to be read by the playback device.

I can not imagine the disaster it would be to use the wrong application on a few thousand tracks, I need some advice.

Thank you for your time.

 

[Vincent Kars]

Replaygain can be found here: https://en.wikipedia.org/wiki/ReplayGain
Indeed a 2001 standard.
EBU R 128 can be found here: https://en.wikipedia.org/wiki/EBU_R_128
Both do the same (loudness normalization) but R128 is considered a better method to cope with the loudness war.

 

[2tec]

Use ReplayGain if your player supports it, as it does not alter the music files, only the tags. Also ReplayGain has both track and album modes. Normalization for when the player does not support ReplayGain. Some forms of volume normalization are one way and cannot be undone; it's therefore good to normalize only a copy of your files.

 

[DVDdoug]

Regular "normalization" is peak normalization (it's a mathematical or statistical concept) and the peaks don't correlate well with perceived loudness. Most commercial music is already peak normalized.

There are a couple of advantages with ReplayGain... (Also MP3Gain & WaveGain).

1. It checks for clipping. If boosting a quiet-sounding song with high peaks (which is not unusual) it won't boost into clipping (unless you change the default settings). Some songs will still remain below the target but they won't be clipped.

BTW - The volume of MOST songs will be lowered and this bothers some people but if you have enough analog gain that's not a problem.)

If you simply Loudness Normalize you'll have to manually check for clipping.

2. It's designed for batch processing. Although if you want to use Album Gain I think you have to scan one album at a time.

...There's one odd thing with ReplayGain (at least the original ReplayGain). It uses an SPL loudness target of 89dB and that correlates to some digital level (which uses a different, but similar, algorithm than UBU R128). An SPL level has to be assumed because of the equal loudness curves. EBU R128 is also assuming some acoustic listening level (unknown to me) but all of the "numbers" are in digital dBFS.

Apple has Sound Check which is similar to ReplayGain.

(my streaming device does support Replay Gain)

All of the popular streaming services are already using their own loudness normalization. But the high-end lossless services may not... It can't be bit-perfect if they are messing with the volume.

And otherwise it shouldn't be possible with streaming because the file has to be scanned in advance....

I can not imagine the disaster it would be to use the wrong application on a few thousand tracks, I need some advice.

It's pretty safe because as you say, it's only adding a "tag" to the file. And you can always disable ReplayGain in your player software.

That's not the case with MP3Gain or WaveGain. They change the actual audio data so they work with any player.

 

[Richx200]

Regular "normalization" is peak normalization (it's a mathematical or statistical concept) and the peaks don't correlate well with perceived loudness. Most commercial music is already peak normalized.

There are a couple of advantages with ReplayGain... (Also MP3Gain & WaveGain).

1. It checks for clipping. If boosting a quiet-sounding song with high peaks (which is not unusual) it won't boost into clipping (unless you change the default settings). Some songs will still remain below the target but they won't be clipped.

BTW - The volume of MOST songs will be lowered and this bothers some people but if you have enough analog gain that's not a problem.)

If you simply Loudness Normalize you'll have to manually check for clipping.

2. It's designed for batch processing. Although if you want to use Album Gain I think you have to scan one album at a time.

...There's one odd thing with ReplayGain (at least the original ReplayGain). It uses an SPL loudness target of 89dB and that correlates to some digital level (which uses a different, but similar, algorithm than UBU R128). An SPL level has to be assumed because of the equal loudness curves. EBU R128 is also assuming some acoustic listening level (unknown to me) but all of the "numbers" are in digital dBFS.

Apple has Sound Check which is similar to ReplayGain.


All of the popular streaming services are already using their own loudness normalization. But the high-end lossless services may not... It can't be bit-perfect if they are messing with the volume.

And otherwise it shouldn't be possible with streaming because the file has to be scanned in advance....


It's pretty safe because as you say, it's only adding a "tag" to the file. And you can always disable ReplayGain in your player software.

That's not the case with MP3Gain or WaveGain. They change the actual audio data so they work with any player.

Thank everyone for your help.

Looks like Replay Gain:

1. I could use a VST for the clipping, and at 100db SPL I still have 20db headroom.
2. The Program I use is PerfectTunes by DbPoweramp it does batch and EBU 128; according to the PertectTunes the track is calculated with EBR128 and a LUFS of -18 target volume.
3, Could always edit the tag or turn off Replay Gain and I use a M4A format.

Thank you for your time

 

[Richx200]

Use ReplayGain if your player supports it, as it does not alter the music files, only the tags. Also ReplayGain has both track and album modes. Normalization for when the player does not support ReplayGain. Some forms of volume normalization are one way and cannot be undone; it's therefore good to normalize only a copy of your files.

Thank you for your time

 

[Richx200]

Replaygain can be found here: https://en.wikipedia.org/wiki/ReplayGain
Indeed a 2001 standard.
EBU R 128 can be found here: https://en.wikipedia.org/wiki/EBU_R_128
Both do the same (loudness normalization) but R128 is considered a better method to cope with the loudness war.

Thank you for your time

Replaygain can be found here: https://en.wikipedia.org/wiki/ReplayGain
Indeed a 2001 standard.
EBU R 128 can be found here: https://en.wikipedia.org/wiki/EBU_R_128
Both do the same (loudness normalization) but R128 is considered a better method to cope with the loudness war.

I appreciate your time

 

[AnalogSteph]

2. The Program I use is PerfectTunes by DbPoweramp it does batch and EBU 128; according to the PertectTunes the track is calculated with EBR128 and a LUFS of -18 target volume.

That's equivalent to what Foobar2000 does. It's basically R.128 scanning aiming for the same loudness as "traditional" ReplayGain for compatibility with older scans still employing the legacy algorithm.

 

[ZolaIII]

That's equivalent to what Foobar2000 does. It's basically R.128 scanning aiming for the same loudness as "traditional" ReplayGain for compatibility with older scans still employing the legacy algorithm.

It's what EBU R128 does plugged in anything.
Again EBU R128 to source and equal loudness compensation (ISO 226) to SPL.
Nice enough explained and without sosages.

https://tech.ebu.ch/files/live/sites/tech/files/shared/tech/tech3343.pdf