[vorbis] ReplayGain support for Vorbis

[Wilson]

----- Original Message -----
From: "Moritz Grimm" <gtgbr at gmx.net>
To: <vorbis at xiph.org>
Sent: Wednesday, January 16, 2002 5:05 AM
Subject: Re: [vorbis] ReplayGain support for Vorbis

Before I begin, I'd like to point out that I understand your points. I just
want to make my perspective clear, in case anyone would like to hear the
other side of the story.

> Wilson wrote:
> > > Ummm... but the point in lossless coding is having the original back
> > > after decoding it. Any alteration to that audio makes no sense to me.
> > > Why don't you use Ogg at -q8 or higher instead (or mp+ --insane, if
you
> > I'm talking about 160GB hard drives here. A single one can hold almost
400
> > CDs in FLAC format.
>
> Oh, and you call those affordable?! For me and propably many others this
> will be the case in 1-2 years, I guess.

You can pick up a 120GB hard drive now for the price of a single tire. I
call that affordable.

>
> Anyways, an affordable 60GB harddisk still has a lot of room, so I
> accept this argument - although I could think of more useful ways to
> fill it up (I ripped my CDs to able to conveniently enjoy them while
> being at the computer, and this didn't make my CDs disappear as the
> perfect archive).

This is straying from the point. My argument is "assume that space is cheap,
as in, no longer a scarce commodity." Assume that you've already managed to
store those "more useful things", and you STILL have enough space left to
store all your CDs as either FLAC or Ogg Vorbis.

>
> > q9 Oggs are coming right up there on FLAC bitrates, so why not take the
> > extra step and remove any possible ugly results? (udial.wav, even though
>
> q9's nominal bitrate is 320kbps while I rarely get a better lossless
> compression than x0.6 (~830kbps, with LPAC which is slightly better than
> FLAC). Besides, what's the point in using Ogg Vorbis in the first place
> if ugly results happen even at q9? udial.wav is no example - there is no
> such thing as clipping in not totally braindead audio-masters. Ogg
> Vorbis is definitely good enough now to be an alternative to lossless
> coding - even for audiophiles (IMHO).

I agree. udial is an awful example. I think it's a bit early to say that Ogg
Vorbis does not fail on any other audio sample, though. Are you suggesting
that no one can tell the difference between ogg q9 and FLAC? I know I can't
right now with my current equipment, but that's not the idea.

>
> > can be certain that you won't "suddenly" start hearing an artifact when
you
> > upgrade your equipment. I'm sure q9 Ogg is good enough, but given the
minor
> > incremental cost of moving up to FLAC, it's an option I would like to
have.
> > Also, AFAIU, you can return to the "original" file with the ReplayGain
> > system. Seems like a good fit with a lossless codec.
>
> Using FLAC instead of Vorbis @ q9 reduces the amount of music on your
> HDD by half or more, IMO not worth the (to me inaudible) gain of
> quality. The noise of your harddisk and fans etc are louder (and
> propably more annoying) than anything that could happen at q9 in Ogg and
> not happen in FLAC.

If you have enough capacity to store all of your music in either FLAC or
Ogg, switching to FLAC does not reduce the amount of music you can store. My
music server is inaudible (fanless.) My "listening" PC can barely be heard
at the bottom of human perception, like a faint whisper. I agree that even
that whisper is probably louder than any Ogg artifact. I'm not really this
picky in "real life", I'm just trying to prepare for the future. I HATE
re-ripping 350 CDs again and again.
>
> I listen to music on a Sennheiser HD 590, and the rest of my equipment,
> while not being high-end, also can be called "Hi-Fi". I don't fear any
> bad surprises, even when I might upgrade somewhen in the far future.

I'm in a similar boat. HD-580s.

>
> > I'm having trouble thinking of a use for an Ogg file that has a Vorbis
AND a
> > FLAC stream. Help me out here?
>
> Like Ogg FLAC, this is a hypothetical example that requires some tools
> to be written first. This is science fiction:

<snip hypothetical situation>

It's not fair to argue that "space is precious!" in the first part of your
message, then come up with an example that requires far more space than FLAC
does in the second half. :)

> To get back to replaygain, currently its values are stored in the Vorbis
> comment field. So, every Vorbis stream in an .ogg file has its own
> replaygain value. A global replaygain for all Vorbis streams and the
> FLAC stream would break the thing - the FLAC mixdown would need
> completely different values than the Vorbis master (in this example),
> and different songs in multiple Vorbis streams would need different
> replaygain values, too. Think of an .ogg file with an English and a
> German track for an Ogg Tarkin videostream, for example.
>
<snip>

I still don't understand why we couldn't store the ReplayGain data in the
metadata stream, and thus have independent support for normalization of
however many audio streams we have in the Ogg container. Can you think of a
reason why this wouldn't work?

<p><p><p>--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.