Hello,

As I understand Ogg files can be converted to lower quality ogg files, without the quality being lower than directly converting from wav to the lower ogg quality. So from wav to ogg quality 7 and then to ogg quality 2 would give the same result as wav to ogg quality 2. Is this true?
Are there programs that will allow me to make the above conversion?
And can you list them in you download area?

Thank you for you'r answer.

I don't think so - in my opinion you should simply avoid transcoding.
Or this his statement true? I like to hear the answer…

Bitrate Peeling in ogg AFAIK is like how 5.1 used to be with mp3, it's in the standards, it just hasn't been done yet (no software, not in the encoder?, i really don't know)

The term bitrate peeling means losslessly transcoding to a lower quality setting, so it is, indeed, what gottkaiser is referring to.

Unfortunately, i know of no formats that currently support it. Doing a search on 'bitrate peeling' in the Ogg Vorbis forums (where, incidentally, this thread would also be more appropriate ) will yield more information. From what i (think i) know, bitrate peeling in Ogg Vorbis has been planned for a long time, but never implemented.

Xiph.org even has a bounty out for it.

Wow... looks like you accidentally created an oxymoron. I mean "losslessly transoding" & "lower quality level" are quite contradictory.

Bitrate peeling is supposed to be a easy'n'quick solution for lowering the bitrate within streaming servers for example. Its quality is likely to be always worse compared to fully decoding and re-encoding at a lower quality level. Though this is a more speculative answer there are several technical reasons which seem to back this theory.

SebastianG

Bitrate peeling was one of the more hyped features of Ogg Vorbis which was supposed to make it easier to make lower quality files by peeling away high qualty ones. Essentially, it would take considerably less time to convert your Vorbis files to a lower bitrate for portables than transcoding would. Unfortunately, it seems good quality bitrate peelers are non-existant and the feature is looking more and more like a "vapour-feature".

Segher did have a functioning bitrate peeler for April fools day a few years ago, though the quality of the peeled files was pretty bad, IIRC. What I dont understand is why Monty didn't seem to have bitrate peeling in mind when he was actually implementing Vorbis. Obviously the idea was out there early on and was touted as a promising feature, but current Vorbis-encoded files are pretty much impossible to peel in an optimal way.

As for 5.1 coupled Vorbis, it looks a bit more promising as Monty did have that in mind and left a comment in the source code saying this is where it would most likely be implemented.

LOL, you're right, i could have made an effort to formulate that better..

What I meant was that the following two procedures would/should result in two streams of the same quality:
- encoding from original source to quality level A, then bitrate peeling to quality level B (which is lower than A)
- encoding from original source to quality level B

Losslessly in the sense that, if you already had level A, bitrate peeling it to quality level B would cause no quality loss comparing to re-encoding to quality level B from scratch.

I was always under the impression that peeling was to be of higher quality than transcoding to lower bitrates, but of (edit: slightly) lower quality than encoding to lower bitrates from the original. (edit: oh, and peeling was to be fast.)

Ok, I got it...
Though, I seriosly doubt that this will ever be the case.

SebastianG

In my mind I have as a project in a near future (maybe one year or two) to put a internet radio station.

For this I had two options in mind:

OPTION 1): Original ripped wavs directly to ogg stream [-q0]
OPTION 2): OGG files (1.1RC1 -q7; obtained from original ripped wavs) to ogg stream [-q0].

I wanted to know if the ogg stream OPTION 1) had the same sound quality than the ogg stream obtained in OPTION 2).

For this; I ripped 10 wavs from original CDs (with EAC); the material was ROCK and JAZZ in general.

I created 10 ogg files [1.1RC1], type OPTION 1): original wavs --> ogg [-q0]
and;
10 more of type OPTION 2): original wavs --> ogg [-q7] --> ogg [-q0]

I compared these files with ABX tests [OPTION 1) vs. OPTION 2)]

The results was not good
I could easy ABX , without a wrong trial.

After, I compared the files OPTION 1) and OPTION 2) with the originals wavs; and, in OPTION 1) the sound quality was a lot better than OPTION 2).

I realized that, in the OPTION 2) the phasing artifacts were more noticeable than in the OPTION 1)

NOTE: I made this experiment two months ago; and I think it is important to notify it now.

On the other hand my intention of this feedback is for positive purposes..

I believe that, in the near future the HA vorbis developers will get the solution to this problem.