Volume normalizing

[sokd623]

Hello,

I just recently started using dBpowerAMP and it seems like it has everything I need for my music collection.
I have about 4000 audio files of different music styles, formats, bitrates, volumes and so on... Some sound very high and other very low.
My goal is to normalize the volume so that all files will play at the same or approximately the same volume level for playback in the car.
I tested Volume Normalizer SDP last night, reducing the volume by -2 DB and it did sound significantly lower than the untouched file and this is great news for me.
However, I don't think I can apply -2 DB on 4000 files that are coded at different volumes. Some will need to be brought down and some will need to be brought up.
I know that ReplayGain does that with metadata, but I prefer resample the files to ensure it always works.

Is there a way to calculate the average volume of all 4000 files with ReplayGain and resample them to the calculated level with Volume Normalizer?
Or, maybe I should be using ReplayGain (Apply) for this purpose? I don't know the difference between ReplayGain and ReplayGaine (Apply)yet.
Also, how do I do it across the board when I have Flac, MP3, M4A, and possibly others? Some albums are associated with tags and they may have files of two or three different types.

Thank you,

[garym]

Re: Volume normalizing

you should use ReplayGain (apply). This will bring all 4000 files to a standard volume, and apply the relative change needed (that is, recreate the audio itself) so that the files will all have a similar volume level without the need for a player/server that can read replaygain tags.

edit: I recommend keeping a copy of your original files, as once you apply the volume normalization, your files are no longer bitperfect copies of the original CDs. And this process is NOT reversible. And so many modern players now support ReplayGain tag info, there will likely be a time in the future that you'll be happy to have the original files with just RG tags added (and NOT volume normalized).

[garym]

Re: Volume normalizing

Also, how do I do it across the board when I have Flac, MP3, M4A, and possibly others? Some albums are associated with tags and they may have files of two or three different types.

It doesn't matter that you have different file types. The process is not comparing your files to each other. They are comparing the loudness of your files, whatever format, to an EXTERNAL standard (e.g., 89db loudness).

[sokd623]

Re: Volume normalizing

Thank you Gary M.

So all files need to be in one single folder or can I load them into the Converter using the top folder and then it reads the content of all subfolders?
Also, do I use only ReplayGain (Apply) or do I have to add Bit Depth and Resample effects, and in what order should they be placed in the DSP effects window?

Bit Depth
Resample
ReplayGain (Apply)

Or differently?

Oh I forgot, what should I set the Encoding to and what ReplayGains options/settings to choose?
There are too many possibilities.

Thank you,

[garym]

Re: Volume normalizing

Thank you Gary M.

So all files need to be in one single folder or can I load them into the Converter using the top folder and then it reads the content of all subfolders?
Also, do I use only ReplayGain (Apply) or do I have to add Bit Depth and Resample effects, and in what order should they be placed in the DSP effects window?

Bit Depth
Resample
ReplayGain (Apply)

Or differently?

Oh and I forgot, what should I set the Encoding to?

Thank you,

you do not need bit depth or resample. You need only replay gain apply. You can put them all in one folder or you can just choose the top folder and it will also work on any of your sub folders underneath that top folder. You tick all this in the batch converter app within dbpoweramp. For encoding select reply gain apply DSP from drop down list.

[sokd623]

Re: Volume normalizing

you do not need bit depth or resample. You need only replay gain apply. You can put them all in one folder or you can just choose the top folder and it will also work on any of your sub folders underneath that top folder. You tick all this in the batch converter app within dbpoweramp. For encoding select reply gain apply DSP from drop down list.

Something doesn't add up...
On the Encoding dropdown list, I only have ReplayGain



Then there are some more options under the Write dropdown list, I chose Track & Album Gain



Then under Advanced Settings there are more options, I chose All Files in Same Folder and I think I need to select ReplayGain instead of EBU R128



All this time the Add button next to the DSP Effects remains grayed out.
What am I doing wrong? How can I add ReplayGain (Apply) DSP?

Thank you,

[garym]

Re: Volume normalizing

Sorry, I've misled you a bit on this (as I don't ever use ReplayGain(apply). I've now looked more carefully and suggest the following steps. There may be a way to combine these steps to simplify (do the RG tagging and applying in one step), but this is how I'd do it:

1. Load up all your files in Batch Converter, and convert to REPLAYGAIN DSP (from the dropdown list). In the options for ReplayGain, I'd choose: Track & Album gain, EBUR128, -18LUFS, and do NOT tick disable clip prevention. (note, EBU R128 is still writing replaygain tags, just a more modern algorithm)

2. At this point you've written ReplayGain tags (metadata) to all your files. The next step will be "applying" the RG info to actually change the audio itself in these files. So you'll be writing new mp3, flac, Alac, etc. files at this point. If you want to convert your mp3 to mp3, flac to flac, etc., then you'll need to separate your files in some way (file explorer search, filtering, etc.)

3. Let's say you have a bunch of mp3 files, you will now do Batch Converter, select CONVERT and choose mp3(lame) from dropdown list. Then choose the level you want (-V2, -V1, -V0, etc.). Note redoing a lossy file (mp3, aac, m4a) will lose quality. It is unlikely noticeable). On the other hand, redoing a lossless file (flac, alac, wav, etc.) will not lose any quality.

After you've chosen what codec you want to convert to, in the popup box, chose ADD to add a DSP. Then add "ReplayGain(apply)" dsp. In the settings for this DSP, choose either Track or Album depending on which RG value you want to use. ALBUM RG values maintains volume differences between songs on the SAME album. TRACK RG values ignores intra-album volume differences and tries to make every song sound the same volume (this used to be called "radio" RG, as that is how one hears songs on broadcast radio....all the same relative volume.

4. on the conversion popup, there is a row for setting where you want these new files written. If you want the new files to have the exact same path (artist/album etc.) and same exact file name, then (a) click on the dropdown menu and select "edit dynamic naming", then select a base location (other than your existing files, as you don't want them overwritten). So if your current files are in c:\users\yourname\music, select something like c:\users\yourname\NEWMUSIC. Then below this, in the dynamic naming box, use this:
[origpath]\[origfilename]

5. When all this is set, bottom right of this popup, select CONVERT.

Obviously test all this on a few files/folders first before you apply to your entire 4,000 files.

[sokd623]

Re: Volume normalizing

Now it makes sense. Running batch on all files to calculate ReplayGain and write tags, then running ReplayGain (Apply) separately on the different file types - that wouldn't matter because the ReplayGain added to all files was calculated on all of them being a part of the same album.

Just one more thing, what do I choose in the Album Identified By menu, All Files in Batch Same Album?

[garym]

Re: Volume normalizing

Now it makes sense. Running batch on all files to calculate ReplayGain and write tags, then running ReplayGain (Apply) separately on the different file types - that wouldn't matter because the ReplayGain added to all files was calculated on all of them being a part of the same album.

Just one more thing, what do I choose in the Album Identified By menu, All Files in Batch Same Album?

[ATTACH=CONFIG]3423[/ATTACH]

Assuming your files all have the metadata ALBUM field completed (name of album), then I'd choose "Album ID Tag"

[garym]

Re: Volume normalizing

Now it makes sense. Running batch on all files to calculate ReplayGain and write tags, then running ReplayGain (Apply) separately on the different file types - that wouldn't matter because the ReplayGain added to all files was calculated on all of them being a part of the same album.

what do you mean it doesn't matter. What format are the current files and what format are you going to make them in this conversion (e.g., mp3, FLAC, etc.). If you're making all of these files the exact same output format (e.g., mp3 or FLAC) then it really doesn't matter what the INPUT file format is.

[sokd623]

Re: Volume normalizing

what do you mean it doesn't matter. What format are the current files and what format are you going to make them in this conversion (e.g., mp3, FLAC, etc.). If you're making all of these files the exact same output format (e.g., mp3 or FLAC) then it really doesn't matter what the INPUT file format is.

I mean it's like you said "The next step will be "applying" the RG info to actually change the audio itself in these files". Once the RG is calculated and written as a tag to all of the files together, I can do RG (Apply) by file type because to my understanding, it kind of "burns" the values in the metadata so they become a part of the file, sorry don't know how else to describe it. That's why I can do it by file type instead of on all of them at the same time. Also, what's MP3 will stay MP3 and what's FLAC will stay FLAC.

That brings me to the next question, I understand that with FLAC, there is no problem, but with MP3, some albums, or content of the same artist, I should say is with different bit rates, when I do the RG (Apply), can I do it on all the files of the same type or do I need to choose the right bit rate as well? For example, can I run RG (Apply) on 128 bit and 196 bit MP3 file at the same time?

[garym]

Re: Volume normalizing

I mean it's like you said "The next step will be "applying" the RG info to actually change the audio itself in these files". Once the RG is calculated and written as a tag to all of the files together, I can do RG (Apply) by file type because to my understanding, it kind of "burns" the values in the metadata so they become a part of the file, sorry don't know how else to describe it. That's why I can do it by file type instead of on all of them at the same time. Also, what's MP3 will stay MP3 and what's FLAC will stay FLAC.

That brings me to the next question, I understand that with FLAC, there is no problem, but with MP3, some albums, or content of the same artist, I should say is with different bit rates, when I do the RG (Apply), can I do it on all the files of the same type or do I need to choose the right bit rate as well? For example, can I run RG (Apply) on 128 bit and 196 bit MP3 file at the same time?

correct. for the first step (adding the ReplayGain tags to the metadata), then you can run this on all your files at once. It doesn't matter if some are mp3 and some are FLAC. It is the "ReplayGain(apply)" step where you are rewriting the actual files when you need to treat mp3, m4a, aac, flac, etc. separately. It doesn't really matter if you separate 128 from 196 mp3. In fact, if I'm rewriting mp3 files (of any bitrate), I'm likely going to convert them to very high bit rate mp3 (maybe -V0, avg bit rate of 240). Here's a bit longer explanation as to why.

For lossless files (FLAC, etc.), you can create new versions of these files as often as you like and the new version will still be a bit perfect version of the old version (except for the volume change for loudness, which means these no longer are bit perfect compared to the CD). But for lossy files (mp3, m4a, aac, etc.) every generation of new file throws out a bit of information. Think of the old days and cassette tapes. If you made a cassette from a vinyl album, it sounded pretty good. If you made a copy of that tape, it's now a 2nd generation, and doesn't sound quite as good, but probably OK. But what if you make a copy of a copy of a copy. At some point, maybe the 5th or 6th generation tape, you'll find that you've lost too much of the sound quality. mp3 files are similar. Each time you create a new mp3 file, you're losing a bit. Now mp3 creation is a LOT better than copying a tape, so you can probably do this a few times and never notice any change in the sound.

But because of the above, I'd like my newly created mp3 files (after ReplayGain(apply) is used) to be a fairly high bit rate mp3, so that I'm not throwing out as much audio info. If it was me, I'd convert all the mp3 at lame -V0. Then again, if it was me I wouldn't do any of this, I'd just make sure I used "replaygain" tag aware players. But say I had a car unit that couldn't handle replaygain tags. In that case, I'd take ALL my files, flac, mp3, etc. and do the procedure I've recommended and use ReplayGain (apply) and make them all, say mp3(lame) -V2 (avg about 192kbps). That's likely good enough. And I wouldn't care because these are my "throwaway" copies that I just use in the car. My main copies are all the original files. And if I get a new car that has a player that can use replaygain tags, then I'll just use my original files and get rid of the "volume adjusted" files I created.

[garym]

Re: Volume normalizing

it kind of "burns" the values in the metadata so they become a part of the file, sorry don't know how else to describe it.

To be clear, it's not just "burning the values into the metadata". This process is permanently CHANGING the actual audio of the file.

[schmidj]

Re: Volume normalizing

Once again, we have a user confused by replay gain. Gary has attempted to provide some guidance, but I'd like to provide some further information.


dBpoweramp provides three different DSP processes that are appropriate for adjusting playback volume. Replay Gain, Replay Gain (apply) and Volume Normalize. Each works differently.

Replay Gain calculates and stores a tag with each track which is essentially a remote volume control for the playback software. But for it to work, the playback software must have been written to read the tag and properly apply it. (I will add here that automobile "radio" players are notorious for being in the 18th century when it comes to reading and properly applying tags, both for selecting tracks to play and for applying things like replay gain adjustments.)

When using the Replay Gain DSP, the original audio data is unchanged. All that is done is to store a tag with the track containing a number which tells the player to turn the volume up or down by some number of decibels.

There are several settings for the Replay Gain DSP which I'll discuss in a future post.


Replay Gain (apply) actually modifies the stored audio data for each selected track, increasing or deceasing the level by the amount previously stored in the Replay Gain tag for that track. Once again, there are a few settings to choose from. Before using the Replay Gain (apply) DSP, one must have stored replay gain tags for the tracks, either by running the Replay Gain DSP or using other software that provides the same function.

It is very important for the user to have saved a duplicate copy of all your tracks before running Replay Gain (apply), as it modifies the stored audio and is irreversible. If you don't like the results, you will have the option to go back to the duplicate copy you saved before running the DSP, without having to rip the CDs again.


Volume Normalize, like Replay Gain (apply), modifies the audio data, but does it in one pass, calculating the necessary modification and applying it to the audio data in the file. It also has many more options for determining how much to adjust the volume. You may in particular find the "adaptive" options of interest for your automobile playback. This appears to be a better option than using Replay Gain and then Replay Gain (apply), but once again, save a backup of the originally ripped files before experimenting with it, this is not reversible.

I've never used Volume Normalize and only used Replay Gain (apply) once for a special project long ago when the settings were different.

Now a suggestion and thought. I'd assume your car "radio" can play music stored on a "smart" phone via Bluetooth unless you are still driving that 1960 Dodge. If your phone has sufficient storage and/or has provision to hold a (now inexpensive) micro SDXC card, put your music on the phone, which will have several well provisioned players available, and use Bluetooth to play the music from the phone.

In my case I have over 110,000 tracks of music on my server, mostly stored as FLAC with Volume Normalize tags. I have m4a copies of them on a one terabyte micro SDXC card in my phone. My car radio is set to automatically pair to the phone when I'm driving. That's where I listen to the majority of my music. Unfortunately, I see less and less phones that make provision for SDXC cards. Could that possibly be from influence from Spotify, Amazon and such which want you to pay them to hear music?

Gary's point about re-encoding mp3 or m4a tracks is well taken, a good reason to rip CDs to FLAC or ALAC as your master copies, and then convert a duplicate copy to mp3 or m4a for more "compact" portable storage and playback. However, I think it depends a lot on the source material and the condition of the listener's ears, and to a lesser degree on the quality of the playback device. Old rock, pop music, I defy you to hear the difference, the damage was done in the recording. In your car, driving along the highway, you'll never hear the difference, having everything the same level is much more desirable. And my almost 80 year old ears have gotten to the point where I probably can't tell anymore. But if you listen to recently recorded jazz or classical music at home on a decent system (not Alexia!) you'd want to avoid too many generations of re-encoding and keep the compressed bit rate up there.

And storing the music on your phone, playing with a replay gain aware player via Bluetooth to your car will avoid any re-encoding.

I'll write a separate piece on some of the replay gain or Volume Normalize settings to think about later.

[garym]

Re: Volume normalizing

Thanks @schmidj. Your clear explanations and detailed background are always much appreciated. I love replaygain function, but it probably is one of the most misunderstood functions around!

edit: your explanation of "volume normalize" vs "replaygain(apply)" will be very useful. (particularly given that I've never done either)