Repl Gain work w. a fold full of various formats? Options

[burn out]

Big Hello to all
I have a large collection of MP3 files but startet to learn and install MPC devices just today. I'm also ooking forward to get prepared for OGG this night.
I'm new to Replay Gain and wonder if I need to install an extra Repl. Gain for each format or if one RG does it for all.
Thanx for pointers in advance

[dev0]

Newbies often get confused by ReplayGain and the different implementations, so I will try to sum up the most important facts/tools:

Introduction
ReplayGain is the name of a technology created to achieve the same playback volume of audio files. It specifies the reference level of 89db and an algorithm to measure the perceived loudness of audio data.

Implementations
There are different ReplayGain implementations, each with its own uses and strength. Most of them use Meta Data to indicate the level of the volume change (Volume is adjusted on playback; needs player/decoder support) others modify the Audio Data itself. Generally it is recommended to use an implementation which uses MetaData and does not touch the audio itself.

WaveGain
Format: PCM Wave
Method: Audio

MP3Gain
Format: MP3
Method: Audio or Meta

VorbisGain
Format: Ogg Vorbis
Method: Meta

replaygain
Format: Musepack
Method: Meta

foobar2000 ReplayGain Scanner
Format: MP3, MPC, Vorbis, APE, FLAC, WAV using APEv2 tags, MP4/AAC
Method: Meta

flac / metaflac
Format: FLAC
Method: Meta

This post has been edited by dev0: Oct 15 2003, 19:51

[picmixer]

Am not sure wether all file formats suppport replaygain natively. That might depend on wether it is implemented in the encoder used and in the decoder plugins for the different players.

AFAIK MPC, Vorbis (Vorbisgain) and FLAC, do support it natively and you can add Replaygain Info directly during encoding.

I personally find the easiest sollution though to use Foobar to ReplayGain all my tracks after encoding.

Foobar 2000 will support Replaygain on all the file formats it can play back and add the Replaygain Info into the files APE tags. It will also add Track Gain and Album Gain simultanously in one go.

EDIT: Just saw that dev0 already explained it way better and with more detail. Seem to be on the slow side today.

This post has been edited by picmixer: Oct 15 2003, 19:52

[burn out]

Thank you f. the good info. It's me who is on the slow side today. Gimme one more try please.
If I got it all right, then I'll have to install an own Replay G. for each format. So if there were OGG, MPC and MP3 that I'm playing out of my collection, there would be 3 different R.Gain programs needed?
Does Foobar replay Gain it w. just one multipurpose device or does it use a bundle of all these known programs?
What I would like to know besides this, since we're at it and probably not in a typical Foobar zone.
What's the deal w. foobar? I wonder if it is an outstanding good player, since it has an own forum in this great site. Is it consumer friendly, so that family men like me may spend more time w. their kids than infront of a computer?
I haven't really been too deep in the FooBar Forum but it looked like a bunch of problems that the user were trieing to handle.
I use WinAmp 2.91 w/o any sign of trouble. I'm not a tweak freak, but I would consider to try Foobar (strange name BTW) if it figures out that it really had significant advantages to offer in sound and ease.
I hope not to stir up things between the fan blocks LOL.
Thanx f. your time

This post has been edited by burn out: Oct 15 2003, 21:24

[spoon]

haven't really been too deep in the FooBar Forum but it looked like a bunch of problems that the user

Never judge a product on its forum, they give the wrong impression. The worst products are the ones without public discussion - just an email address.

[Garf]

If I got it all right, then I'll have to install an own Replay G. for each format. So if there were OGG, MPC and MP3 that I'm playing out of my collection, there would be 3 different R.Gain programs needed?

Yes.

Does Foobar replay Gain it w. just one multipurpose device or does it use a bundle of all these known programs?

Foobar has 1 replaygain scanner and supports writing replaygain information to the formats that support it. For formats that don't, foobar stores the replaygain values in an internal database.

What's the deal w. foobar? I wonder if it is an outstanding good player, since it has an own forum in this great site. Is it consumer friendly, so that family men like me may spend more time w. their kids than infront of a computer?

It has a very basic (read:non-bloated) interface but is very powerfull. Whether you consider that consumer friendly is very personal.

I haven't really been too deep in the FooBar Forum but it looked like a bunch of problems that the user were trieing to handle.

The statisfied people you don't hear

I use WinAmp 2.91 w/o any sign of trouble. I'm not a tweak freak, but I would consider to try Foobar (strange name BTW) if it figures out that it really had significant advantages to offer in sound and ease.

Replaygain support is definetely much better in foobar. Also things like gapless playback work better. You can try it and see if you like it.

[picmixer]

Does Foobar replay Gain it w. just one multipurpose device or does it use a bundle of all these known programs?

Indeed, Foobar does it's replaygain scanning with one device, it's replaygain scanner. It works on all formats and will write the RG info into the file tag (EDIT:or as Garf already said, into it's database). Obviously it will depend on your Playback Device or Software Player wether it will work properly during playback. Obviously Foobar wil support them all during playback. Am not sure exactly wich Winamp inputs suport Replay Gain. I am pretty sure it works for FLAC and MPC. Not sure about the rest though.

What's the deal w. foobar? I wonder if it is an outstanding good player, since it has an own forum in this great site. Is it consumer friendly, so that family men like me may spend more time w. their kids than infront of a computer?

I guess the deal is (IMHO), that it's an Audio Player with many nifty features. Except for secure ripping it does everything for me that I need to do with my Audio files. One of it's outstanding features is for example excellent ReplayGain support (trying to stay on topic).

I haven't really been too deep in the FooBar Forum but it looked like a bunch of problems that the user were trieing to handle.

Ah no, we're just a bunch of slightly to demanding users that like to keep the developers busy . And after all thats what fora are partly there for, to help people solve their problems (at least the ones considering software,etc.).

Have to admit that FB2K is quite customizable, but it should work just fine out of the box. Noo need to go into to much tweaking, unless you really want to. And if thats the case I am sure you'll always find someone willing to help out in said Forum.

I'd say, just give it a try and see if you like it. It might not be love at first sight, but I am warning you the day might come when you never want to revert to another player anymore .

In case you don't like it, you can always go back to using Winamp (2.9x) , no harm done. Fine player as well and in the end it all comes down to personal preferences.

EDIT: To slow again

This post has been edited by picmixer: Oct 15 2003, 22:33

[Prodoc]

How about the quality of the different ReplayGain programs?
I'm not realy interested in speed but what can you tell me about the accuracy of the different programs?
Will e.g. Foobar be enough to use to determine the ReplayGain values for the different audio formats (mp3 and ogg)? Or am I better of using VorbisGain for my ogg files and e.g. MP3Gain for my mp3 files?

[dev0]

foobar's Replaygain component, flac's Replaygain routines and Wavegain's ones are all based on the vorbisgain sources, so they should give similiar/equal results.
Frank Klemm did some tuning of the algorithm for Musepacks's implementation, but it should be safe to use too.
MP3Gain modifying the actual audio data is limited to 1.5db steps.

This post has been edited by dev0: Oct 16 2003, 17:59

[ErikS]

foobar's Replaygain component, flac's Replaygain routines and Wavegain's ones are all based on the vorbisgain sources, so they should give similiar/equal results. steps.

They all use they old version you say? Amazing...

[Garf]

foobar's Replaygain component, flac's Replaygain routines and Wavegain's ones are all based on the vorbisgain sources, so they should give similiar/equal results. steps.

They all use they old version you say? Amazing...

Why not? It's not like there's something wrong with it...

[ErikS]

Why not? It's not like there's something wrong with it...

Well, why use something which is slightly worse (even if it's ever so little) when there is a better alternative that wouldn't cost more than a couple of minutes to integrate?

[Garf]

Why not? It's not like there's something wrong with it...

Well, why use something which is slightly worse (even if it's ever so little) when there is a better alternative that wouldn't cost more than a couple of minutes to integrate?

Who said anything about the old implementation being worse?

[ErikS]

Why not? It's not like there's something wrong with it...

Well, why use something which is slightly worse (even if it's ever so little) when there is a better alternative that wouldn't cost more than a couple of minutes to integrate?

Who said anything about the old implementation being worse?

I have some faint memories of Frank mentioning improvements in the context of the new replaygain. That usually implies that the other party is worse somehow... Want quotes?

[Garf]

I have some faint memories of Frank mentioning improvements in the context of the new replaygain. That usually implies that the other party is worse somehow... Want quotes?

The 'tweaks' were never tested to be a general improvement. I'm not going to change anything away from the reference code (which works extremely well) until it can be shown that the new code a) has no failure cases b) is an actual improvement.

[ErikS]

The 'tweaks' were never tested to be a general improvement. I'm not going to change anything away from the reference code (which works extremely well) until it can be shown that the new code a) has no failure cases B) is an actual improvement.

A is a tough one. But possibly I can dig out something for b. I'll notify you in case I find anything...

[Ardax]

MP3Gain modifying the actual audio data is limited to 0.5db steps.

[nitpick] 1.5db. [/nitpick]

[dev0]

*fixed*

[2Bdecided]

There is at least one other "improved" version floating around - it uses hanning windowing and 50% overlap to improve bass accuracy (in certain cases), and adds some speed increases elsewhere to still manage an overall speed increase (despite doing twice as much work).

Frank did one improvement (time smearing, I think), and then commented out these lines in his code, at least in the version that I checked (and I'm not a programmer, so I could be wrong). He has also added something that prevents you amplifying silence by 80dB.


Replay Gain has the following "faults":
The frequency compensation is too simplistic, and fails for some extreme tracks.
Very short sounds aren't perceived with the correct loudness.
Exeptionally dynamic tracks are sometimes difficult to loudness match (not a fault, just a fact).
A spectrally dense signal can sound louder than a spectrally thin signal of similar amplitude, but replay gain doesn't "know" this. It rarely causes a problem with rael music, but you can make test files to show how big the problem is.

An accurate basilar membrane kind of simulation (like in Moore's and Zwicker's work) would work better, but someone who understands psychoacoustics _and_ C/C++ coding would need to implement it, otherwise it would be desperately slow.

Cheers,
David.

[Peter]

Somewhat offtopic note, I've managed to significantly speed up foobar2000's replaygain scanner (originally based on vorbisgain algorithm), using assembly language (and some 3dnow, but the 3dnow part doesn't matter that much), sourcecode of all changes is available. I'm planning to convert it to some sort of portable non-environment-specific library soon.

[Peter]

Some random info:
- I'll have replaygain scanner library that is portable and doesn't contain any OS/platform-specific functions done later this week, BSD license, to be included with foobar2000 SDK.
- Menno generated more tables for sample rates that weren't supported by original vorbisgain (192khz, 96khz, 37.8khz, etc).
- Apparently SSRCing signal (on unsupported samplerates) before calculating gain is accurate enough (+-0.01dB gain change on some test file, rounding errors can give same difference).
- I have bad headache today

[sv3n]

I have a ReplayGain question that has to do with my Rio Karma. I wasn't sure whether I should ask it on Riovolution.com or here, but because the people here are generally more knowledgeable about ReplayGain, I decided this would be the best place:

I have a collection of MP3 and Vorbis files. I ripped all of the Vorbis files myself, while the MP3’s were obtained… using other methods... .
The problem is that the Karma does not yet support ReplayGain tags in any format. I’ve used MP3Gain to balance all of my MP3’s at 89.0 dB, but it seems that all of my Vorbis files are noticeably louder than that.

How can I get all of these files balanced at the same volume on my player, without causing clipping by going above 89 dB or damaging them by re-encoding?

Thanks!