Ogg -q4 transparent to you? Options

[aabxx]

Considering quality 0 sounds very good to me, I would reckon something between quality 1 and quality 2 would be enough for me >90% of the time. Let's say quality 2 to be on the safe side. So never mind quality 4

[CioCio]

-q2 drives me crazy trying to ABX, because pretty soon everything starts to sound the same. I have my collection at -q4 and now I'm debating going down to -q2 or -q2.5

It'd save me several gigs, but I might be getting a new hard drive soon anyways, so I don't know

[pepoluan]

CioCio: Just keep your collection as it is, unless you are *really* pressed for space.

But next time you encode into Vorbis: Now you know what -q to use

[CioCio]

CioCio: Just keep your collection as it is, unless you are *really* pressed for space.

But next time you encode into Vorbis: Now you know what -q to use

Haha, thanks for the advice! Yeah, I wish I hadn't been so lazy before doing it originally

[hybridfan]

I use -q5 normally and that sounds very good in my shell likes, but yes I would say that -q4 does sound tranparent for me.

[HeXeR]

I usually listen to Rock and Metal, sometimes Pop, and I was able to ABX it up to -q3. It wasn't even a killer sample, they were some tracks from my own collection. At -q4 everything was abrupt transparent to me, so this is what I use.

[Whelkman]

I echo the sentiments of many in saying I'm surprised any codec could sound as indistinguishable from source at such low bitrates. I am a victim of an outlier case, unfortunately, in that Vorbis/aoTuV Beta 5 does not handle stereo chiptunes at all well, specifically dual POKEY audio.

For the uninitiated, POKEY is an audio/controller chip found in virtually every Atari computer and arcade machine from the late 70s to the mid 80s; it produces primitive square waves and white noise across four channels. A common (and cheap--$5!) modification to the XE line of home computers was to solder a second POKEY chip for pseudo stereo sound. These days, one can best listen to these tunes using a POKEY emulator, such as ASAP.

I suppose this setup creates a worst case scenario for Vorbis stereo coupling as some tunes leverage eight possible channels by playing completely different waveforms in left and right. The end result is that, for this type of music, Vorbis is essentially useless below -q 6.0. I can reliably ABX at -q 5.0 with cheap headphones and a blaring air conditioner less than five feet to my right; by -q 5.9 I think I still hear a difference but can't (patiently) sort it out.

By contrast, I can't distinguish at all between source and LAME -V6 --vbr-new. In fact, the 12,100 Hz drop with -V7 is more transparent to me than Vorbis at -q 5.0. On my testing sample (full song), the bitrate difference between Vorbis -q 6.0 and LAME -V6 --vbr-new is 224 - 136 = 88. Wow, that represents an additional 65% overhead for a comparative level of transparency! In this scenario I'm less disappointed in Vorbis than I am impressed with LAME. Lossy encoders once sucked at encoding primitive waveforms, but now I can achieve casual transparency on tough, "unrealistic" stereo samples at a mere 128-150 kbps.

For the record, I'm not the habitual ABXing type, but when I started experimenting once again with Vorbis against this particular type of music I immediately knew something was amiss. This is the most extensive testing I've ever conducted; I just couldn't believe Vorbis could perform so badly against anything compared to MP3. I guess there's still work to be done, though I doubt this type of composition is a high priority for tuning. In fairness, Vorbis becomes the clear winner in the 64-80 kbps range; here LAME becomes garbled while Vorbis merely sounds a little muddled--and not too different from results 100 kbps higher, strangely.

Samples are available if anyone's interested.

This post has been edited by Whelkman: Jul 31 2007, 07:12

[slks]

I'm interested. Upload these samples.

[Whelkman]

http://www.hydrogenaudio.org/forums/index....showtopic=56591

This post has been edited by Whelkman: Aug 4 2007, 02:44

[Bourne]

This test is the most crazy thing I have gone through concerning ABXing audio. Truly pointless!
To me, LAME V6 is beaten by itself at V4 or V2. OGG is pretty transparent at -q4 already, would easily be equivalent to Lame V0.

This post has been edited by Bourne: Aug 28 2007, 06:59

[shadowking]

This test is the most crazy thing I have gone through concerning ABXing audio. Truly pointless!
To me, LAME V6 is beaten by itself at V4 or V2. OGG is pretty transparent at -q4 already, would easily be equivalent to Lame V0.

Hmm.. Maybe on those mp3 bad samples, Otherwise I really doubt this claim that 130k ogg = 230k mp3. On average the two are close at 128k. Guruboolez still rated -V2 anchor higher than all codecs in his 128k test of over 100 samples. In the last two public tests, LAME -V5 is competitive with ogg Q4. If you want to push it then ogg q4 *might* compete with LAME -V4.

This post has been edited by shadowking: Aug 28 2007, 07:41

[Aoyumi]

BTW: with the new b5, I can see (by no way hear) low-level noise in the spectrum at about 20kHz (with 1kHz bandwidth) when using q5 that did not appear when encoding to q5 with aoTuV r1. What is the cause of this?

I have already revised the problem. However, the official release of the version including it has not yet come.

[Ragnarok]

I agree on the most part, -q4 weather abxing or not they generaly sounds transparent to my ears ( baring killer samples of course) whatever equipment I listen on which included a ogg album decoded burnt cd-r vs the original on a tag maclaren setup.

However the artifacts in ogg are very diffrent, not really anoying or very noticable. I can pick out some some samples in q3 it's also for the odd brief moment their is a partiular sound that you notice a drop in calarity as if the sound for that partiular frequency had is sample rate droped compaired to a uncompressed/lossless or q4 version.

OGG vorbis is a great format. Still for serious listening I'd rather have a lossless or uncompressed version.

[Raptus]

Managed to ABX some non-killer samples at q5, so I settled with q6.

[Raptus]

Trust me: For personal use and enjoyment, ABX-ing is totally useless. You'll never be able to enjoy your songs again, always worrying that it is non-transparently encoded.

Well, then I must be the exception
I've done extensive ABXing, I've become very sensitive to artifacts, yet I settled for a lossy codec but at a quality that is transparent to me on all _music_ I tested. I never looked back and have no problem simply enjoying the music.
Yes, I occasionally notice artifacts in the original recording (bad edits, noise filtering or samples taken from bad mp3's!), but I'm not bothered, I just note it. It only starts to become annoying when I listen to songs I know encoded by other people and start hearing artifacts I'm sure aren't in the original

[naylor83]

I've done extensive ABXing, I've become very sensitive to artifacts, yet I settled for a lossy codec but at a quality that is transparent to me on all _music_ I tested. I never looked back and have no problem simply enjoying the music.

I couldn't agree more. By properly testing beforehand at which Q setting I could detect artifacts in a number of my tunes, I now feel great knowing that I'm using the minimum amount of disk space possible for a compressed and (to me) transparent archive of my music.

[pepoluan]

Trust me: For personal use and enjoyment, ABX-ing is totally useless. You'll never be able to enjoy your songs again, always worrying that it is non-transparently encoded.

Well, then I must be the exception
I've done extensive ABXing, I've become very sensitive to artifacts, yet I settled for a lossy codec but at a quality that is transparent to me on all _music_ I tested. I never looked back and have no problem simply enjoying the music.
Yes, I occasionally notice artifacts in the original recording (bad edits, noise filtering or samples taken from bad mp3's!), but I'm not bothered, I just note it. It only starts to become annoying when I listen to songs I know encoded by other people and start hearing artifacts I'm sure aren't in the original

Ah, I may be a bit too harsh in stating that

Anyways, I have to add that there are 2 kinds of people:
- People who are annoyed by artifacts
- People who are not (so) annoyed by artifacts

My comments apply mostly to those of the first kind.

I myself fall squarely within the definition of the second kind. And I think, you, too

[naylor83]

Umm.... I'm definitely annoyed by artifacts. That's why I abxed and came to the conclusion that I can't hear artifacts using Ogg Vorbis Q4.

[Francksoy-Wizzu]

Hello all.

This whole topic is what finally made me register, after all these months spent in reading the Hydrogenaudio forums.

My initial, fast ABX'ing with pop music made me believe that Q1 was good enough for me for mobile purposes. 'Great', I though, 'I can can fit so much music in my 2GB Sansa this way'. Well, I was wrong.

After I started being annoyed by artifacts mainly on vocal classical music, further and more careful ABX'ing and repeated listening taught me that my personal comfort zone starts at Q3 (112Kbps) and that Q4 is, like for many others, mostly transparent to my ears. Some very rare tracks are not 100% OK for me at this setting, but I don't care that much for mobile use.

But yes, I think that Q4 (128Kbps) is mostly transparent, and can probably be even 100% transparent to some less picky/fussy ears than mine. Sometimes I feel like hearing all these small details is a curse.

[stevenv]

After I started being annoyed by artifacts mainly on vocal classical music ...

Hi,

It would be great if you could give (ex/s)amples where this is the case at "-q 1".

thnx,
Steven

[dsimcha]

Hi,

I just started transcoding my FLAC albums (I'll save them) to Ogg. I first tested q 5, q 6 and then q 4 and I can't tell any difference between them.

But I cant believe that 128kbit files are absolutely transparent? I know it's very subjective, but I wanted to ask you if it's also transparent to you or if you hear any artifacts?

Greets

Edit: I use AoTuV beta 5

I've come to the conclusion that "absolute" transparency is not the best benchmark. If you demand perfection and have plenty of space, go with lossless. It's guaranteed to be transparent even on the most killer samples you can throw at it. Using aoTuv Beta 5.7, it seems to me like the drop-off in ABXable samples with increasing quality is roughly exponential. It seems like increasing the quality setting by one unit makes about half of the samples that I could previously ABX transparent. I could probably ABX about 50% of my music at -q1, 25% at -q2, 12.5% at -q3, etc. I use -q2 for my portable player, as I've never found a sample that's easy to ABX and a lot of samples are transparent. I feel anything more would take up a lot of extra space for very little additional quality. However, I've found at least one killer sample in music I actually listen to, not in a search specifically for killer samples, that I could ABX all the way up to -q7. Therefore, on my PC, where space is ridiculously cheap and plentiful, I use FLAC.

[RogerG]

-q4 is intransparent to me using aoTuV b5.7 even with my crappy old hardware setup. I'd never use something below q6. I need a big buffer because in the future I am going to have a better soundcard and headphones.