Volume Leveling and Winamp

[Aminifu]

PART 1

Why Volume Leveling is Needed

Although music is encoded to a digital format with a clearly defined maximum peak amplitude, and although most recordings are normalized to utilize this peak amplitude, not all recordings sound equally loud. This is because once this peak amplitude is reached, perceived loudness can be further increased through signal-processing techniques such as dynamic range compression and equalization. Therefore, the loudness of a given album has more to do with the year of issue or the whim of the producer than anything else. Because of this, a random play through a music collection can have significant volume changes with every other track.

There is a solution to this annoyance: within each audio file, information can be stored about what volume change would be required to play each track or album at a standard loudness, and music players can use this "replay gain" information to automatically nudge the volume up or down as required.

What is ReplayGain

The ReplayGain specification is a standard which defines an appropriate reference level, explains a way of calculating and representing the ideal replay gain for a given track or album, and provides guidance for players to make the required volume adjustment during playback. The standard also specifies a means to prevent clipping when the calculated replay gain exceeds the limits of digital audio, and it describes how the replay gain information is stored within audio files.

The audio industry does not have a standard for playback system calibration, but in the movie industry a calibration standard has been defined by the Society of Motion Picture and Television Engineers (SMPTE). The standard states that a single channel pink noise signal with an RMS level of -20 dB relative to a full-scale sinusoid should be reproduced at 83 dB SPL.

ReplayGain adapts the SMPTE calibration concept for music playback. Under ReplayGain, audio is played so that its loudness matches the loudness of a pink noise signal with an RMS level of -14 dB relative to a full-scale sinusoid.

In ReplayGain implementations, the reference level is described in terms of the SMPTE SPL playback level. By the SMPTE definition, the 83 dB SPL reference corresponds to -20FS dB system headroom. The -14 dB headroom used by ReplayGain therefore corresponds to an 89 dB SPL playback level on a SMPTE calibrated system and so is said to be operating with an 89 dB reference level.

SMPTE cinema calibration calls for a single channel of pink noise reproduced through a single loudspeaker. In music applications, the ideal level of the music is actually the loudness when two speakers are in use. So, ReplayGain is calibrated to two channels of pink noise.

[Aminifu]

PART 2

ReplayGain and Winamp

The WA implementation of ReplayGain is enabled/disabled via the General Preferences - Playback - Replay Gain options. To apply Replay Gain to your music files, right-click items in the Media Library or Playlist and select 'Calculate Replay Gain' via the 'Send To' menu. The Replay Gain values are stored in the file tags. These tags can only be understood by music players that recognize this particular implementation. The actual audio data is unchanged. Offically supported formats are MP3, WMA, WMV, AAC, MP4, M4A, OGG & FLAC. The wavpack format (wv) is also supported if the wavpack decoder (included in the Winamp Essentials Package) is installed.

The other 8 options on the General Preferences page are;

Amplification Mode: Default selection is 'Apply Gain / Prevent Clipping' (recommended by WA devs). This will level out the volume and prevent distortion. Three other choices are available.

Preferred Source: Choose between 'Track' (default) or 'Album' modes.

Disable if preferred source not available: Disables RG if e.g. 'Album' mode is selected but only 'Track' info is available.

Adjustment for files without RG: Pre-amp selection applied to all files without ReplayGain info (default value is -6.0 dB).

Replay Gain Preamp: Pre-amp selection to adjust global gain for all files with or without RG info applied (default value is 0 dB).

Ask before writing Replay Gain to analyzed files: When check marked, the Replay Gain Analyzer results dialog will appear at the end of each scan. WA dev's recommended preference is to use 'Save as Album', which also writes Track data. When unchecked, values will be saved automatically without prompting.

---Ask after each album is scanned: Prompts to save values after each album is scanned.

---Ask after all files are scanned: Prompts to save values after all files are scanned.


Within the media library's CD ripping options is another ReplayGain related option, "Automatically calculate Replay Gain". This option enables/disables the automatic calculation and application of ReplayGain values to ripped files. This is a hidden process which occurs at the end of each individual track. The default behaviour is to apply Album Gain (which also applies Track Gain). Note that enabling this option may considerably slow down the overall ripping process.


When selected, the WA implementation displays a window that shows the progress of the analysis and the final adjustment(s) that will be applied. However, the results of the analysis are never shown, thus the level of adjustment is not known. In other words, it can not be determined if the final adjustment values are minimum, maximum, or somewhere in between. The final overall volume could be lower than it needs to be, but there is no way to tell.


Undo is simple for officially supported formats. Just use the WA tag editor to delete the ReplayGain tags. A 3rd party tagger must be used to delete ReplayGain tags in a wavpack file and may also be used for the other file formats. Most of the popular 3rd party taggers support wavpack.

[Aminifu]

PART 3

A 3rd Party Alternative for MP3s Only

"MP3Gain" can analyze and adjust mp3 files so that they have the same volume. It is an implementation of the ReplayGain spec. MP3Gain uses David Robinson's Replay Gain algorithm to calculate how loud the file actually sounds to a human's ears. With MP3Gain, the loudness adjustment is done on the data itself, in a lossless/reversible way. Another difference with MP3Gain is the fact that it can only adjust physical volume in 1.5 dB steps. There is no quality lost in the change because the app adjusts the mp3 file directly, without decoding and re-encoding.

Here's the technical reason on why it's lossless (despite operating on the data itself), and also why the smallest change possible is 1.5 dB:

The MP3 format stores the sound information in small chunks called "frames". Each frame represents a fraction of a second of sound. In each frame there is a "global gain" field. This field holds an 8-bit integer which can represent values from 0 to 255.

When an MP3 player decodes the sound in the frame, it uses the global gain field to multiply the decoded sound samples by 2^(gain/4). If you add 1 to this field in all the MP3 frames, you effectively multiply the amplitude of the whole file by 2^(1/4) = 119 % = +1.5 dB. Likewise, if you subtract 1 from this field, you multiply the amplitude by 2^(-1/4) = 84 % = -1.5 dB, reducing the amplitude of the whole file.


The way MP3Gain works actually has a very strong benefit. Since it is the data itself that is modified, MP3Gain does not require special support from music players.


MP3Gain can store 'analysis' and 'undo' information in special tags inside the mp3 file itself. These tags are in the APEv2 format. If you choose not to use these tags or delete them, then you will not be able to automatically undo changes made by MP3Gain. You will still be able to undo any changes, but you will have to manually keep track of what changes where made to your files and undo the changes manually. Logs of the analysis, changes, and errors can be saved to text files with names and locations of your choosing. MP3Gain must be used to undo any changes made with MP3Gain.


After MP3Gain completes an analysis, the results are displayed with an indication of whether clipping is present. Adjustments can then be chosen (up or down) before they are applied, to see their effect on clipping. This allows a minimum adjustment to be made when the main goal is to prevent clipping. Of course, the same adjustment (minimum or not) needs to be made to all files when the main goal is overall volume leveling.


CAUTION!!!

There is another app with a similar name. "Mp3Gain PRO" does a form of volume normalization inside the mp3s. If you feel a song is too quiet or too loud at the beginning (or middle, or end), then it can boost or reduce the volume just for that part. Pretty cool, if that's what you need. In order to make its fine-tuned adjustments, it must re-encode the mp3 file. The changes Mp3Gain PRO makes are not undo-able. For this reason, I do not recommend it for general use.

[Aminifu]

I have used both WA's and MP3Gain's implementations of ReplayGain for years without any apparent damage to my music files.

MP3Gain usually allows me to achieve volume leveling at a higher overall level than WA's ReplayGain, but it only works on mp3s. In addition, WA's lower level is easily dealt with by kicking the master volume control up a notch or three.

My main issue with WA's implementation is that it does not work with some my portable music players.

Has anyone found other methods of volume leveling (and/or clipping prevention) that work well with WA, other than 'normalizing' the wav files before trans-coding (which is my least favorite option since it tends to smother the music's dynamics)?

[MrSinatra]

i don't think most people know what volume leveling is or means, so i'd say replay gain instead, as that is actually mentioned in the prefs and other areas of the app/tags, unlike "volume leveling." (they won't know what RG is either, but at least they might have seen it)

you can also apply RG via the center top file info window's buttons, but this is not best practices, since if you have multiple tracks from the same album, you need to do them all at once to get the right album RG level. so anytime multiple tracks from one album are being done, they should all be done together to get the right album RG setting. in addition, album tags must be present, (and maybe some combo of other tags depending of different factors).

you say "the results of the analysis are never shown." well, first they are in the tags, and in the ML columns if selected. but those analysis results can be shown if they are applied manually, so you can see them prior to application.

also, i do not believe winamp allows the user to remove just RG values, at least from id3. you'd have to delete the whole tag. a better undo is to just deactivate the RG feature in winamp, or use mp3tag to delete the RG values.

i would never use mp3gain b/c i do not want to impose or use APE tags on my files, and i would not want to need them to undo changes. imo, id3 works best with mp3, APE causes trouble.

mp3gain IS useful for situations where the user must adjust the data itself, b/c normal RG is not supported, but in such a case, i would adjust my hardware/software first, before i started applying mp3gain and APE to all my files. there's no question that i have DL'd mp3s that had mp3gain applied and i am stuck with them that way b/c the ape tags are gone.

you could also explain that all RG does is change the volume, but it changes it across the whole track. and the other tag it uses, peak value, is what it uses to prevent clipping.

[Aminifu]

i don't think most people know what volume leveling is or means, so i'd say replay gain instead, as that is actually mentioned in the prefs and other areas of the app/tags, unlike "volume leveling." (they won't know what RG is either, but at least they might have seen it)

you could also explain that all RG does is change the volume, but it changes it across the whole track. and the other tag it uses, peak value, is what it uses to prevent clipping.

you say "the results of the analysis are never shown." well, first they are in the tags, and in the ML columns if selected. but those analysis results can be shown if they are applied manually, so you can see them prior to application.

I can come up with a better title, equating volume leveling and RG. How about:
Volume Leveling/ReplayGain (Keeping all your tracks at close to the same volume)
Too long?

I though the text made it clear that all RG does is change the volume across the whole track. I can try to make it clearer. RG is a technique. The way it is applied by WA and mp3gain are different. The bytes that control volume across the entire track are adjusted real time by WA's implementation. MP3Gain adjusts these bytes directly, so there is no need to adjust them during playback. The algorithms that WA and mp3gain use to determine these adjustments are probably close, but not the same (patent issues). The one mp3gain uses was stated, don't know what WA uses. I can also add more info on how clipping prevention is done, but again both implementations use the same techniques to accomplish it.

As to RG analysis results, WA does show 1 interim result, the peak value detected (if, like you say, the user elects to see it). This result is also shown on the advanced tab of the WA tag editor for flac files, but it is not shown in the WA tag editor for mp3 and wavepack files. Don't know about other formats, but unless they use vorbis tags I don't think it will be shown for them either. Likewise, the peak value is not one of the column choices in the media library. MP3Gain will provide more interim analysis results than the peak value detected, but as with WA, the user has to elect to see them. Also WA doesn't tell you if clipping was detected, mp3gain does.

Both implementations show the bottom line, namely the album and/or track values that will be used to control volume (including clipping prevention, if needed). These are the final analysis results, so I should have wrote that WA only shows 1 interim analysis result and mp3gain shows more.

you can also apply RG via the center top file info window's buttons, but this is not best practices, since if you have multiple tracks from the same album, you need to do them all at once to get the right album RG level. so anytime multiple tracks from one album are being done, they should all be done together to get the right album RG setting. in addition, album tags must be present, (and maybe some combo of other tags depending of different factors).

You are referring to the default Big Bento skin. I don't use that and didn't know. But since it is possible for skins to provide alternate ways to apply WA's RG, that needs to be mentioned, along with the drawbacks you stated.

Can you point me to where I may find the info to clear up the doubt about tag combos.

also, i do not believe winamp allows the user to remove just RG values, at least from id3. you'd have to delete the whole tag. a better undo is to just deactivate the RG feature in winamp, or use mp3tag to delete the RG values.

WA's RG adjustment values can be removed independent of the other tags for flac files. I assume this would be true for other formats that use vorbis tags, but they can not be removed independently for formats that use id3 tags. Your other points here are valid and should be added.

i would never use mp3gain b/c i do not want to impose or use APE tags on my files, and i would not want to need them to undo changes. imo, id3 works best with mp3, APE causes trouble.

mp3gain IS useful for situations where the user must adjust the data itself, b/c normal RG is not supported, but in such a case, i would adjust my hardware/software first, before i started applying mp3gain and APE to all my files. there's no question that i have DL'd mp3s that had mp3gain applied and i am stuck with them that way b/c the ape tags are gone.

Is the APE tag trouble in general or due to bugs in WA?

True, if you don't use the APE tag to record the mp3gain change, you can not use mp3gain to automatically return the track to the original volume level. But you can keep a log of the change and manually use mp3gain to reverse the change.

Also true, if you don't use the APE tag or keep a log or otherwise don't know what the change was, you can't get back to the original volume level. But you are not stuck. You can still use mp3gain to make another change, up or down, to the overall volume level.

This brings up another issue I did not mention. In the case where you need to increase the overall volume, this is awkward with WA's RG implementation. You have to use the pre-amp adjustment. This then affects all the other files with established WA RG values, since this adjustment is applied ahead of the WA RG adjustments. If you need to use the pre-amp for some files, then I think you would need to reanalyze all the files you did without this pre-amp adjustment being applied to prevent them becoming too loud again.


Ok, thanx for your input. Now how do I apply all these changes? Will a moderator let me back into the original posts or do I have to do them all over again?

[MrSinatra]

I can come up with a better title, equating volume leveling and RG. How about:
Volume Leveling/ReplayGain (Keeping all your tracks at close to the same volume)
Too long?

i think thats good. i might say: "Volume Leveling/ReplayGain (Playing back your tracks at similar volume levels automatically)"

there probably also needs to be a clear explanation in the text of "gain" basically meaning "volume"

I though the text made it clear that all RG does is change the volume across the whole track. I can try to make it clearer. RG is a technique. The way it is applied by WA and mp3gain are different.

personally, i think RG and mp3gain should not be conflated. they have the same goal, but they are otherwise completely different, and mp3gain really has nothing to do with winamp specifically. i think there should very clear delineation between them, totally different methods.

i also forgot to ask if mp3gain can support an "album" mode? i think it doesn't, but i could be wrong. if not, another big strike in my mind.

might be an idea to explain that "track gain" is best for any and all listening EXCEPT when listening to any two or more tracks off of the same album in sequential order. in that case, "album gain" is the better option to preserve the intended differences in dynamic range, (like DSOTM has for instance)

The bytes that control volume across the entire track are adjusted real time by WA's implementation. MP3Gain adjusts these bytes directly, so there is no need to adjust them during playback.

i think winamp just reads the tag and adjusts the gain directly to the winamp volume slider. so if the slider is at 75%, it does that plus 3db if thats what the tag says. but i'm just guessing.

The algorithms that WA and mp3gain use to determine these adjustments are probably close, but not the same (patent issues). The one mp3gain uses was stated, don't know what WA uses. I can also add more info on how clipping prevention is done, but again both implementations use the same techniques to accomplish it.

that doesn't really concern me. meaning, how they arrive at their values isn't really pertinent imo. what is pertinent, is how it works. so like one is done on tag reading playback, the other direct to the audio, that kind of thing.

As to RG analysis results, WA does show 1 interim result, the peak value detected (if, like you say, the user elects to see it).

i'm not sure what you are talking about here? if i have RG tags set to be manually applied, i can see both track and album RG values and peak values PRIOR to application. thats the default.

personally, i don't need to see them, so i make it so it just writes everything automatically.

This result is also shown on the advanced tab of the WA tag editor for flac files, but it is not shown in the WA tag editor for mp3 and wavepack files.

well, i said that once written, you can't edit the id3 ones anyway, and you probably can't see peak values for id3 at all. but u can see the RG values in the tags and ML. (another reason to use vorbis over id3 btw where possible)

Don't know about other formats, but unless they use vorbis tags I don't think it will be shown for them either. Likewise, the peak value is not one of the column choices in the media library. MP3Gain will provide more interim analysis results than the peak value detected, but as with WA, the user has to elect to see them. Also WA doesn't tell you if clipping was detected, mp3gain does.

no, the default is you will be shown them in winamp, and its all values. u have to elect NOT to see them, (and actually, i think that should be the default)

Both implementations show the bottom line, namely the album and/or track values that will be used to control volume (including clipping prevention, if needed). These are the final analysis results, so I should have wrote that WA only shows 1 interim analysis result and mp3gain shows more.

i'm really not following you, and i'm not sure if its because you're mistaken about things, or i don't understand the terms you're using.

You are referring to the default Big Bento skin. I don't use that and didn't know. But since it is possible for skins to provide alternate ways to apply WA's RG, that needs to be mentioned, along with the drawbacks you stated.

well, esp since its the default.

Can you point me to where I may find the info to clear up the doubt about tag combos.

not sure what you mean.

WA's RG adjustment values can be removed independent of the other tags for flac files. I assume this would be true for other formats that use vorbis tags, but they can not be removed independently for formats that use id3 tags. Your other points here are valid and should be added.

umm, thats exactly what i said. i specifically said id3.

Is the APE tag trouble in general or due to bugs in WA?

imo, they're trouble. we've had this discussion before, just recently in fact, and i pointed you to threads about APE and the problems i had with them.

imo, best to strip them, and not deal with them. and that in turn, is another strike against mp3gain. and i would also point out that 1.5db increments are a bit too far for my tastes. dbs are created as the levels the human ear can detect differences between, so one would want increments of one db or less, not more than one db.

True, if you don't use the APE tag to record the mp3gain change, you can not use mp3gain to automatically return the track to the original volume level. But you can keep a log of the change and manually use mp3gain to reverse the change.

how many people do you expect would do this? its like saying use a phone book instead of google, or actually WRITE the phone book.

Also true, if you don't use the APE tag or keep a log or otherwise don't know what the change was, you can't get back to the original volume level. But you are not stuck. You can still use mp3gain to make another change, up or down, to the overall volume level.

and how would you know what the orig volume level is to get back to?

This brings up another issue I did not mention. In the case where you need to increase the overall volume, this is awkward with WA's RG implementation. You have to use the pre-amp adjustment. This then affects all the other files with established WA RG values, since this adjustment is applied ahead of the WA RG adjustments. If you need to use the pre-amp for some files, then I think you would need to reanalyze all the files you did without this pre-amp adjustment being applied to prevent them becoming too loud again.

i'm not following you here at all.

all the pre-amp does is adjust the baseline. so if ones RG tags get the files to 89db, and i adjust the pre-amp to +2 db, then now all the RG files will playback at 91db. where's the problem in that?

the only issue is the preamp adjust ALL files, including ones WITHOUT tags, but you can compensate for that by adjusting down the db setting for files WITHOUT RG tags, which by default is already at -6db.

Ok, thanx for your input. Now how do I apply all these changes? Will a moderator let me back into the original posts or do I have to do them all over again?

you probably need to do them again. i would start a new thread with new copy, and add a link from this one to the new one.

[mindbytez.com]

RAC - a live leveler (gain control)

Maybe you check out a Winamp-plugin I made, which levels all sound live while playing. Version 1.1 controls the gain in the mid frequencies for better listening pleasure.

The plugin is stuck in the Winamp-plugin-review for months now. Somebody still doing this for Winamp? I even made a new version since then, version 1.1. You can download it from the site at

http://mindbytez.com

On the site go to the "RAC".

The plugin is best suited for movies, but it works on music too. Play around with the settings if you like.

Richard

[MrSinatra]

interesting, how does it work? compression? buffers? does it adjust gain midway thru a song? i'd not be into that. but for some internet streams, wavs and so on, i could see a use.

[Aminifu]

Hi MrSinatra,

I tried to respond to your post #7, but took too long. The token expired. Will try again next week when I have more free time.

[Aminifu]

Maybe you check out a Winamp-plugin I made, which levels all sound live while playing. Version 1.1 controls the gain in the mid frequencies for better listening pleasure.

The plugin is best suited for movies, but it works on music too. Play around with the settings if you like.

Richard

Hi Richard,

I see your plug-in is a DSP. Does it provide linking to another DSP or is a stacker required?

[mindbytez.com]

Gain Control Plug-in

It is a Winamp Nullsoft Signal Processing Studio DSP plugin. I do not know a way to stack those or link to another DSP.

If there is a way, please point me to it so I can code it in.

[DrO]

you cannot directly do what Aminifu is mentioning (linking to another DSP) if this is just a sps file - the only way would be to use a DSP stacker plug-in to load the Nullsoft SPS plug-in with your sps file active and the other DSP plug-in(s) required.

and this is stuck in review as there were issues with the review system where submissions would not show up (from around February to April) and as no one on the team has confirmed if it is fixed or not (despite asking), i've now given up doing plug-in reviews so i've no idea who is doing plug-in reviews now (i assume no one).

-daz

[Aminifu]

Hi MrSinatra,

This response to your post # 7 is a summary rather than in a 'quote - reply' format since so many points are made in that post. If I miss something you would like me to address, let me know.


WA RG and mp3gain are not conflated, imo. The article is written in 3 parts to keep things separate. Part 1 discusses the general issue of tracks playing at different volume levels and the replay gain specification developed to deal with the issue. Part 2 discusses WA's implementation of the replay gain spec. Part 3 discusses a separate app that applies the replay gain spec for mp3s only.

MP3Gain does support album mode. The difference between album and track modes was not explained. Thanx for pointing out that omission. Personally I only use track mode. Some users may want to maintain the relative difference in volume between tracks on an album, so that info needs to be added.

I agree that many would not care about how things work, I included the technical stuff (including the comments on the difference in the reported analysis results) for those who would be interested. As to the WA analysis results, I was talking about the detected peak value not being shown in the media library and it's not shown in the WA tag editor, except for tracks with vorbis tags. The rest of the analysis stuff I was talking about could be 'overkill', since it makes no difference to the bottom line.

The pertinent point I should add is that the WA RG implementation does not allow a target level to be selected and the mp3gain RG implementation does. Some have complained that the target level WA's RG uses is too low. Careful use of the pre-amp adjustment could deal with this, but too large an adjustment could cause clipping. Some discussion of why clipping is undesirable probably needs to be added to part 1 of the article.


The 1.5db step increment/decrement is a function of the mp3 file format. This is explained in the article. Both WA and mp3gain have this limitation for mp3s. I have not been able to find out what step resolution WA's RG can achieve with other formats.


Before switching to mp3gain, I don't remember WA's RG needing the album tag or any other combinations of tags to be present in order to analyze and add replay gain tags. MP3gain does not have this restriction. I need to do some testing of this.


As far as returning to the original volume level. I was incomplete/wrong in regard to WA. Except for the WA tag editor being able to remove RG tags for formats that use vorgis tags, a 3rd party tagger is needed to remove these tags for other formats or WA can be 'told' to ignore the RG tags, like you say.

For mp3gain, the APE tags it can insert are needed to have mp3gain automatically return tracks to their original levels. Or, mp3gain can be 'told' to create the logs containing the info needed to use mp3gain to return tracks to their original levels manually. This is explained in the article. If you download a file that was 'adjusted' by mp3gain and you don't have the APE tags or the logs, then you can't get back to the original level, but you can still change the volume level of the mp3 further, up or down as needed.

Personally, I don't care about returning tracks to their original levels, since the whole point for me is having all tracks (even those from the same album) have close to the same overall volume level. The info about returning to the original levels, for both WA and mp3gain, is provided for those who would be concerned about this.


The example you gave for a track with low overall volume would probably not be a problem. But what about a track that needs a much larger adjustment. Then the other files, with RG tags, would be increased by this large adjustment also, maybe into clipping range. The only 'safe' way I see would be to re-analyze, but I'm not sure the pre-amp level is taken into account with WA's RG implementation. I need to do some testing of this also.


In places I agreed with the points you made and simply rephrased in my own words. Thank you for your comments. They have helped me realize stuff I had overlooked.


The article is not for new users only, hopefully those with some experience would get something out of it too.

[Aminifu]

I came across the article linked below that has an interesting take on loudness levels in new CDs of new and old music.

http://www.howtogeek.com/trivia/the-phenomenon-of-mastering-music-with-increasing-loudness-is-known-as-what/