Help - Search - Members - Calendar
Full Version: Ogg development speed...
Hydrogenaudio Forums > Lossy Audio Compression > Ogg Vorbis > Ogg Vorbis - General
Pages: 1, 2
gnoshi
mmm... wholesale re-encoding.
Been waiting for this day for a long time =)
(well, waiting for the day when the listening tests on ogg vorbis 1.0 are done and the consensus is something along the lines of 'yeah cool go for it u fiends', but close)

Finally a good reason to go flush out those beta 4 and rc2 files (got rid of the beta 2 and beta 3 ones a while back =)

gnoshi - a happy puppy for the day
kxy
The things that I care the most is that the -q 6 will be lossless instead of q5. And you can't put a hold on the channel coupling, because the ECTL call does not exist right now.

Yea, now would be a good time for listening test. cool.gif
mithrandir
QUOTE
Originally posted by john33
@mithrandir

No, you'd be wrong!wink.gif  Numerous changes have been made. I would suggest re-encoding would be a worthwhile exercise, but you may like to test a few tracks before undertaking a wholesale re-encoding!smile.gif

Really? I mean, the pre-RC4 encoder is only a week old and look how long it took to go from RC2 to RC3 to pre-RC4. I don't keep up with the Vorbis development teams workload, but logically I thought they couldn't do much to the psymodel and stuff in a week.

Anyway, I think pre-RC4 sounds pretty damn nice at low bitrates so while reencoding to 1.0 means I'll have to redo about 30 albums, if it's even better quality perhaps it's worth it.
Ardax
QUOTE
Originally posted by rjamorim
Well, I would believe now is the right time for some serious listening tests. smile.gif


Ahem. We've already started. smile.gif
rjamorim
QUOTE
Originally posted by Ardax
Ahem.  We've already started.  smile.gif


Nah. I didn't meant finding problem samples. I was thinking of s full ABC/HR test, conduced by our friend ff123, with some samples at low bitrates (64kbps). Or medium bitrates (128kbps)...
indybrett
Well, since I'm not up for doing listening tests and ABX'ing and all that stuff, I decided to go another route.

I thought it would be interesting to compare file sizes.

I chose an album that I knew (from past experience) would produce bloated files. I encoded Metallica's Load, using LAME 3.92 --alt-preset standard.

Using this method, the whole CD totaled 122MB after encoding. My next step was to see what Ogg quality level would produce similar file sizes.

Using Ogg 1.0 -q7, the entire CD totaled 112MB.

Not very scientific, but if q7 (or anything less) sounds as good or better than APS, then we have progress on the quality vs. file size front.

Guess I should encode an acoustic album and see how it fares with file sizes.

I'll leave the listening tests to the folks with the golden ears.
gnoshi
Hey, just a question for everyone else who has tried 1.0 (I am in the office and haven't poked around much yet).
What quality # correspond to about what bitrates these days? (eg. q 5 used to create about 160kbps files) and at what point does lossless channel coupling click in these days (if it has changed).
Thnx =)

edit: kxy, where did you find out q6 rather than q5 is lossless coupling now? (desparate for info.. =)

gnoshi
tangent
QUOTE
Originally posted by treech
I hope there will be some super-duper ASM optimized version of oggenc after the 1.0 rls (like GOGO), or something like that.

Speed is always a good thing :=)


Yup, but would you call it Gogoogg or Ogggogo?
indybrett
Ogg-a-gogo
treech
oga-oga would be a cool name biggrin.gif biggrin.gif biggrin.gif
rc55
What about Oggasm? tongue.gif

(Ogg/Assembly)

Ruairi
kxy
QUOTE
Originally posted by gnoshi

edit: kxy, where did you find out q6 rather than q5 is lossless coupling now? (desparate for info.. =)

gnoshi


I had a chat with monty online. I was trying to find a quick and dirty hack to enable chanl coupling without regards to bitrates/q level. But he told me that ECTL call basically didn't exist yet. As it requires more testing.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.