Vorbis - Oct 2002 - How to fit Oggs in a specific amount of space?

I took 5 albums (Classical music) and converted them to Ogg Vorbis 
at "Full Bitrate" (-q10) and all 5 directories take up about 775 Megs 
which won't fit on a CD.  So I ripped them again in WAV first (And 
give my friend back his CDs) but now I wanna know what quality 
setting should I use to fit them on 1 CD (The highest possible with 
total space used just under 700 Megs)

        Is there a tool available that will analyse WAV files and determine 
the "Nominal Deviation Factor" or something like that to figure out 
what quality setting I need to use to fit them in a specific amount 
of space.  Is this even consistant?  By that I mean if a certain song 
encoded at q4 (128k nom.) gets an average of 120kbps does that mean 
the same song encoded at q8 (256k nom.) will get an average of 
250kbps?  Or if it's not linear is there a way to predict the q 
setting to use.

        So for files that take up 775 Megs @ q10 to fit in 700 Megs I need 
to use what?  q9?  q9.5?  q9.7?  q9.8?

        There has to be a faster way than trial and error...

+--------------------------------------------------+
|      Martin Fontaine     mfont@videotron.ca      |
+--------------------------------------------------+
| The Braids Bonanza: http://braids.cjb.net        |
| Mel D.'s Good Gifts: http://come.to/meldoane     |
+--------------------------------------------------+
| "Love means facing all your fears." -Amy Sky     |
+--------------------------------------------------+

--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

By the time you typed this up and sent it on its way, you could have
graphed the major points :)

Honestly, wouldn't different music types would compress to different
sizes, so I wouldn't think you can get a one-size-fits-all lookup table.

<p><5 minutes of comparing music length to file size later>

Well, more and less difference than I thought. Depends on the artist
more than the genre, I guess.
Most music at q5.0 is about 1.05mb per 60 seconds, but I can tag a few
of the more "busy" artists and watch them consistently go to 1.2mb,
1.3mb per 60 seconds.

-----Original Message-----
From: owner-vorbis@xiph.org [mailto:owner-vorbis@xiph.org] On Behalf Of
Martin Fontaine
Sent: Monday, October 21, 2002 6:13 PM
To: vorbis@xiph.org
Subject: [vorbis] How to fit Oggs in a specific amount of space?

<p>        I took 5 albums (Classical music) and converted them to Ogg
Vorbis 
at "Full Bitrate" (-q10) and all 5 directories take up about 775 Megs 
which won't fit on a CD.  So I ripped them again in WAV first (And 
give my friend back his CDs) but now I wanna know what quality 
setting should I use to fit them on 1 CD (The highest possible with 
total space used just under 700 Megs)

        Is there a tool available that will analyse WAV files and
determine 
the "Nominal Deviation Factor" or something like that to figure out 
what quality setting I need to use to fit them in a specific amount 
of space.  Is this even consistant?  By that I mean if a certain song 
encoded at q4 (128k nom.) gets an average of 120kbps does that mean 
the same song encoded at q8 (256k nom.) will get an average of 
250kbps?  Or if it's not linear is there a way to predict the q 
setting to use.

<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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

Martin Fontaine wrote:
>         I took 5 albums (Classical music) and converted them to Ogg Vorbis
> at "Full Bitrate" (-q10) and all 5 directories take up about 775
Megs
> which won't fit on a CD.  So I ripped them again in WAV first (And
> give my friend back his CDs) but now I wanna know what quality
> setting should I use to fit them on 1 CD (The highest possible with
> total space used just under 700 Megs)
This is hard to say, since Ogg Vorbis is a true VBR CODEC, it will never
give you any guarantees on bitrates and filesizes. At the expense of
quality (I assume you don't want that, since you're already picking this
IMHO insane quality level -q 10) you can encode the .wav files in
--managed mode and the -b parameters. This allows you to make some
educated guesses on the resulting filesize, because the encoder will
stick closer to the bitrate you pick via -b. Example:

oggenc --managed -b 320 music.wav -o music.ogg

music.wav is 4:30 minutes / 270 seconds long, 320kbps means 320,000 bits
per second, or 40,000 bytes per second. 40,000 * 270 = 10,800,000 bytes,
or 10,546.875 kB, or ca. 10.3 MB.

Note that this is still no absolute value - using --managed mode results
in ABR encoding (Average Bit Rate), so your mileage will still vary.
However, that's something you can work with. --managed -b 320 will sound
roughly the same like -q 9 (i.e. better than -q 9 on easy to encode
music, relatively equal on average music, and it will perform worse on
hard to encode music - that's what you get for restricting the encoder,
you get Variable Quality).

You should be able to figure out the highest bitrate for your music,
which takes n seconds to play, to fit on a CD by yourself. Leave some
headroom to deal with ABR overhead (however, I believe classical music
is among the easy to encode stuff).

Or, you could as well just pick -q 9 ... that might fit well on your CD.
(That is, do you actually hear a difference between -q 9 and -q 10? If
yes, I wouldn't use Ogg Vorbis but something lossless instead.)
> of space.  Is this even consistant?  By that I mean if a certain song
> encoded at q4 (128k nom.) gets an average of 120kbps does that mean
> the same song encoded at q8 (256k nom.) will get an average of
> 250kbps?  Or if it's not linear is there a way to predict the q
> setting to use.
Well, the point of a quality setting is to select the quality *your
ears* need to be happy. While doing so, the encoder picks as many bits
as necessary and as little as possible to maintain the quality you
selected. It depends entirely on the music that you encode, because VBR
doesn't "think bitrate" anywhere. About the linearity - no, it is
not.
At least I don't think it is, from what I've seen so far ... ;P

<p>Moritz
--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

Hello Martin,
  I would try to run them through rehuff. You usually get a 10%
  reduction in file size without loosing any quality. The reason is
  that for encoding vorbis uses some Huffman codes. But since the
  codes must go in the header and the encoder doesn't knows what the
  output will be untill it finishes the file, vorbisenc uses some
  fixed codes that have an almost unmeasurable probability of being
  optimal. Rehuff simply reads the data, computes the optimal codes
  and resave the data using the new codes. Note that this is a
  looseless process.
  BTW, Seher, have you keep polishing rehuff? Will Xiph use it as an
  option for vorbisenc?

Best regards,
Alejandro Belluscio

Monday, October 21, 2002, 8:13:05 PM, you wrote:

MF>         I took 5 albums (Classical music) and converted them to Ogg
Vorbis
MF> at "Full Bitrate" (-q10) and all 5 directories take up about
775 Megs
MF> which won't fit on a CD.  So I ripped them again in WAV first (And 
MF> give my friend back his CDs) but now I wanna know what quality 
MF> setting should I use to fit them on 1 CD (The highest possible with 
MF> total space used just under 700 Megs)

MF>         Is there a tool available that will analyse WAV files and
determine
MF> the "Nominal Deviation Factor" or something like that to figure
out
MF> what quality setting I need to use to fit them in a specific amount 
MF> of space.  Is this even consistant?  By that I mean if a certain song 
MF> encoded at q4 (128k nom.) gets an average of 120kbps does that mean 
MF> the same song encoded at q8 (256k nom.) will get an average of 
MF> 250kbps?  Or if it's not linear is there a way to predict the q 
MF> setting to use.

MF>         So for files that take up 775 Megs @ q10 to fit in 700 Megs I
need
MF> to use what?  q9?  q9.5?  q9.7?  q9.8?

MF>         There has to be a faster way than trial and error...

<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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.