View or see audio levels

[hello_hello]

Ill experiment to see if ffmpeg can add/change the REPLAYGAIN_TRACK_GAIN type metadata.

Do you mean change it to something manually? If so, don't do that. I doubt ffmpeg would offer that ability anyway, as the system can only work if there's a single target volume, which is set in stone for ReplayGain. You can change the physical volume for MP3 or AAC audio, or re-encode to a particular volume, but the ReplayGain tags should always contain a volume relative to 89dB (-18 LUFS). You could edit the ReplayGain fields with Mp3Tag easily enough, but it's a bad idea.

I increased vol by 9db in mp3ProTrim, and reduced it by 9db in a second mp3, then applied Foobars Track Replay Gain. Which If I understand it correctly, doesn't change anything but writes data in the tag area that's used by the player to adjust levels. (Existing tags are preserved.).
FB very accurately reversed my level changes.. (see below, before and after) but seems to set the max level to somewhat less than optimum. A 3db increase would be a good thing. From Googling ffmpeg can adjust levels but I can't find the correct command. Does anyone know it. Ideally the track gain figures from FB could be used to set this ? My goal is to have the mp3 level changed, so no input from the player is needed.

If you save the tags and foobar2000 is adjusting the volume on playback, you can set how it's adjusted in preferences. The settings in the first screenshot below would tell fb2k to adjust the volume on playback to 6dB lower than the usual target volume. So it'd be 83dB or -24 LUFS. Unless an MP3 has a peak that'd still exceed 0dB, in which case the volume would be reduced further.

As support for obeying ReplayGain tags is fairly non-existent in hardware players, I disable fb2k's ability to adjust the volume on playback and adjust it losslessly, or adjust it when re-encoding. With the ability to adjust the volume on playback disabled, when you play an MP3 you're also hearing the "real" volume, and not the ReplayGain adjusted volume.

The second screenshot has fb2k configured to physically adjust the volume of the MP3s losslessly to 83dB, or -24 LUFS. If you physically adjust the volume to 83dB, the ReplayGain track data saved to tags will be updated to +6dB (or the nearest 1.5dB increment), as it's always relative to the target volume of 89dB. If you mess with the tag data, the system can't work as expected.

Even if you've adjusted the volume losslessly to 83dB, you'd still have to configure fb2k as per the first screenshot, or it'll increase the volume to 89dB on playback anyway. Either that, or disable the ReplayGain adjustment completely for playback, as per the third screenshot, or remove the ReplayGain data from the tags, so fb2k would play the MP3s at their new volume. No ReplayGain data in tags is fine. Incorrect ReplayGain data is bad.

It is possible to get fb2k to display the TrackGain/AlbumGain volumes as LUFS, and the peaks as dB, given the ReplayGain method isn't very intuitive (see the 4th screenshot). You can create playlist columns for that. The way it's done in the 4th screenshot requires a plugin and either way, requires a bit of scripting, but I can explain how to do it if you're interested. You might want to give yourself a crash course in fb2k's layout editing first though.

Mp3Gain can also adjust the volume losslessly and save ReplayGain data to tags and it displays before and after clipping info. It has an option to adjust the volume by a specified amount. You can change the "target volume" when losslessly adjusting files, but the ReplayGain tags are still saved or updated to a volume relative to 89dB. It uses the original ReplayGain scanner, which is pretty good, but not quite as accurate as the R128 scanner fb2k uses, and about 100x slower. The data is saved to Ape tags, but unlike fb2k, it includes "undo" information so it's possible to reverse any physical volume changes, as long as the tags remain.
Fb2k is aware of Mp3Gain's Ape tags and will remove the ReplayGain info when scanning or adjusting the volume and save the new info to ID3 tags (by default). Mp3Gain is oblivious to the ReplayGain info saved to ID3 tags though, so if you've previously scanned with fb2k, after adjusting the volume with MP3Gain, the existing info saved by fb2k will be wrong and there'll be conflicting ReplayGain info, so you should remove the ReplayGain info saved by fb2k first. Just something to be aware of.

[Tomycooler]

Hi, I have a bunch of mp3's at various levels and want to make the 'same'.
However I don't want to a program to 'normalise' or change them in any way... just get
an idea what their level is and then manually + or - in Mp3pro Trim as needed.

Is there anything that might do this? Torrent TurboTax Gogoanime
Otherwise I can batch load 20 or so at a time into Audition, but it'd be tedious as there's several hundred.
Thanks

to get a relative indication of how many frames will be affected by the +/- setting.
The mean volume below is -18.5 but has roughly 25000 out of 11823104 total that may be distorted if raised too high.

[kirkmc]

Thanks for all the info/data. A few things above my head (I'm not familiar with LUFS) and to ponder.
>The second screenshot has fb2k configured to physically adjust the volume of the MP3s losslessly
That was interesting.. you're saying fb2k will adjust mp3 volume without re-encoding ?
If, so that may change what I'm doing.. which is to get fb2k Track Gain and Peak Gain with my own VBA script and use that for later adjustment (I can only do now manually in mp3ProTrim but want to eventually automate that).
Desired 'right' level is 0db as shown in the waveshape grid in Abode Audition. Not sure how that equates to the 83db you mention. I load in half a dozen mp3's and compare what's seen to fb2k replay gain figures, looking to devise a 'good enough' conversion formula. (I'm not there yet). I also did much the same thing with ffmpeg's Summary, but hard to know which may be best. fb2k is the easiest.

In the meantime I'll try out your settings/examples. I'd be interested in anything further you have.
Cheers.

[hello_hello]

Thanks for all the info/data. A few things above my head (I'm not familiar with LUFS) and to ponder.
>The second screenshot has fb2k configured to physically adjust the volume of the MP3s losslessly
That was interesting.. you're saying fb2k will adjust mp3 volume without re-encoding ?

Yes. It'll adjust both MP3 and AAC losslessly, and while MP3s would often be standalone files, it'll adjust both when they're in an M4A/MP4 or MKV container. It might support other containers but there's no official list I could find. MKV and M4A are just the two I tested.

If, so that may change what I'm doing.. which is to get fb2k Track Gain and Peak Gain with my own VBA script and use that for later adjustment (I can only do now manually in mp3ProTrim but want to eventually automate that).
Desired 'right' level is 0db as shown in the waveshape grid in Abode Audition. Not sure how that equates to the 83db you mention. I load in half a dozen mp3's and compare what's seen to fb2k replay gain figures, looking to devise a 'good enough' conversion formula. (I'm not there yet). I also did much the same thing with ffmpeg's Summary, but hard to know which may be best. fb2k is the easiest.

In the meantime I'll try out your settings/examples. I'd be interested in anything further you have.
Cheers.

LUFS = loudness units relative to full scale. I think it comes from the EBU R128 standard, but it's interchangeable with dB, so -20dB and -20LUFS are the same thing.

ReplayGain's official target volume is 89dB, but it's actually a measure of sound pressure level based on a SMPTE standard so it's not very intuitive. After translation to human-speak, it's -18dB or -18LUFS. It's sort of like an "average" volume, but in respect to how loud something sounds to our ears. It works pretty well. ReplayGain has it's own spec for scanning the audio and determining the volume, but the newer R128 scanning method is a little better, so while foobar2000 still refers to ReplayGain and it's target volume etc, it actually uses the R128 scanning method.
(FYI, the norm for soundtrack audio is -23dB or -23LUFS).

When you referred to the correct level being 0dB, I assume that's "peak normalising" (ie the loudest peak at 0dB)? The average volume and the loudest peak don't have much of a relationship though. You could peak normalise two files and one might sound much quieter than the other because it has more dynamic range (average volume to peak volume ratio). That's one of the problems ReplayGain/R128 scanning tries to overcome, aiming for "how loud it sounds" instead.

Fb2k doesn't have a peak normalising option as such, but it's still easy enough to do. ReplayGain can be configured for losslessly adjusting the volume, adjusting while converting, and for playback, and the options are in different places. For losslessy adjusting it's under "Preferences/Tools/ReplayGain Scanner/Alter File Content". For converting it's configured as part of a converter setup, and for playback it's in "Preferences/Playback". See the attached screenshots, but don't forget to disable ReplayGain for playback if you want fb2k to play the MP3s at their "real" volume after they've been scanned or adjusted.
The first screenshot is for losslessly adjusting and the second is for adjusting while converting, but they both tell fb2k to adjust the volume to 20dB higher than the standard ReplayGain volume, or until the loudest peak hits maximum, and as the loudest peak will always hit maximum first, the end result is the same as "peak normalising". It does somewhat defeat the purpose of ReplayGain though.... to have them all at the same volume according to your ears.

If you're fussed about the peaks, enable the "True Peak Sampling" option under "Preferences/ReplayGain" and choose the "auto 4x overscanning" option. Without oversampling, the scanner just looks at the sample values. There's an easy to follow explanation of "true peak" here.
https://www.masteringthemix.com/blogs/learn/inter-sample-and-true-peak-metering
Personally, I don't use it much as it slows the scanning speed quite a bit, I don't think I've ever heard the "true peaks" cause clipping, and because the loudest peak for most CD tracks is a fair way below 0dB after the over-all volume is adjusted to -18LUFS (89dB for ReplayGain).

By the way, I was fiddling with my fb2k configuration a few days ago, trying to make the ReplayGain info a bit more accessible, and to display it in human-speak, and after rediscovering fb2k's ability to duplicate "playlist views" it evolved into a tabbed element dedicated to displaying the volume. If you're interested, there's a screenshot here and instructions for creating the required playlist columns in the post above. Those columns can be added to the standard playist easily enough. To have them display the way they are in the screenshot will require a bit of layout editing though.

"Peak normalising" with foobar2000.
Screenshot 1: "Peak normalising" MP3 and AAC losslessly.
Screenshot 2: "Peak normalising" while converting..

[hello_hello]

In the "learn something new every day" department, I was looking for something else and discovered foobar2000 has the ability to edit ReplayGain data. It's a bad idea, but it can be enabled under the right click ReplayGain menu as per the screenshot below.

There's also an option to adjust the volume of MP3 and AAC losslessly by a specified amount. The menu item isn't enabled by default, so if I knew fb2k could do that, I'd forgotten about it. If you adjust the volume losslessly by a specified amount and there's existing ReplayGain info in tags, fb2k updates it so it stays relative to 89dB, which is a good thing.

[kirkmc]

Hi I've set Tools-RG Scanner-After File Content and ContextMnu-RG-Apply custom Gain as per your screenshots.
Then loaded some files and "Apply track replay Gain to file content"

This is all that's need for lossless volume adjust (i.e I'm not missing anything?)

Results are quite good, and I think ok to achieve a better volume "match" with a batch of files. I'll attach A/B screen shots.
I also tried a batch I'd adjusted manually and there was almost no change, which was as hoped.
I didn't apply "True Peak Sampling" as tehre's seems no need, results were fine as is.
And there is no change to tags, which is also good.

Many thanks for all the detail.

[kirkmc]

Screenshots A/B

Track 2 and 6 appear louder than I'd set by hand, but that isn't too obvious just listening. And certainly better than original.