Apple should have gone with Ogg Vorbis

[swhite]

Using a frequency analysis program (Spectrogram), as well as my own ears, I've discovered that Ogg Vorbis sounds far better than AAC and also has way more treble. AAC at 128kbps has a sharp cutoff at 15.8KHz; absolutely NOTHING comes through above that point. Ogg Vorbis has a sharp cutoff at 20.7KHz, but that's still plenty high enough to sound really good. 112kbps Ogg has a cutoff at 17.9KHz, but that's still better than AAC!

Shayne

[Alltaken]

i think lossy formats should be banned LOL :p

Alltaken

[Jay]

[moved to GD as it has really nothing to do with winamp]

[Drunken Master]

wma is really the only way to go.

[taylormemer]

Originally posted by Alltaken
i think lossy formats should be banned LOL :p

Alltaken

In some ways yes but no I don't think so.

[Drunken Master]

Now that I've thought about it, mp3pro is really the best format available.

[DJ Egg]

nah, Real Audio rocks! :D

[Drunken Master]

DJ Egg has a good point.

[Jay]

you guys suck, da best is and always will be midi.

[Drunken Master]

Sure midi is great..... if you like porno music.

[anubis2003]

ah hell no - old Nintendo game formats had the best sounding music. Super Mario Brothers for instance. Dr. Mario is my personal favorite.

[InvisableMan]

apple sucks and is dying as a whole, the ipod cannot save them.

[DJ Egg]

DSP Group TrueSpeech (8 kHz, 1-bit, Mono, 1 kbps) :up: :D

[ScorLibran]

You're all wrong. Yamaha VQF is teh 1337n3ss 4udi0 f0rm4t.




(BTW...the lead post in this thread is *so* getting posted on Hydrogenaudio. It's outstanding! This is the very kind of humor they'd appreciate. :D )

[taylormemer]

I have heaps of stuff in VQF, intresting, I didn't know it was to l33t.:cool:

[DJ Egg]

*cough* Seriously though, yeah Ogg is probably the best compressed lossy format, close call with Musepack MPC maybe? But, to be honest, I couldn't give a damn about what Apple does/uses/says/(anything). Though I'm certainly glad they didn't choose ogg, that's for sure (not that they could/would've anyway). Hmm, Ogg Vorbis audio in an MP4 container . . . most ponderous.

When most people have got 500Gb hard drives or something, from an audiophile point of view, this (http://rarewares.hydrogenaudio.org/lossless.html) will be the way to go.
Hmm, a 10 Terrabyte HD please, yeah, the Western Digital 512mb cache, 57600rpm one please. Thank you.

I did say "seriously" didn't I ?
Oh yeah, I did . . . heh :D

Then again, I guess by then that you'll be able to carry your pc around with you, like a wristwatch or something, heyhey.

Ok, fun's over, time for me to get back to some serious tech support . . . :o :igor: :D

[ScorLibran]

Disclaimer...My apologies of anyone took my last post seriously. It was intended humorously.

For the results of double-blind sound quality tests, check out this 128kbps Listening Test (http://www.hydrogenaudio.org/index.php?showtopic=11936&) conducted late last year.

These four are statistically tied for first place, but...
MPC > AAC > WMA Pro (not normal WMA) > Vorbis

LAME MP3 came in second, behind the formats above, then Blade MP3 was the low anchor.

Neither VQF nor RA are seriously considered high-end encoding formats for reaching transparency (sound quality = to source) at similar bitrates.

Hope this clarifies things.

[Cognition]

Pure wave format. .wav is lossless, especially if you remember to select "Telephone Quality" when recording (using sndrec32 of course). ;)

[ScorLibran]

:weird:

"Telephone quality" PCM WAV...

Nice! ;)

[taylormemer]

Intresting.

[penvzilaa]

Originally posted by swhite
Using a frequency analysis program (Spectrogram), as well as my own ears, I've discovered that Ogg Vorbis sounds far better than AAC and also has way more treble. AAC at 128kbps has a sharp cutoff at 15.8KHz; absolutely NOTHING comes through above that point. Ogg Vorbis has a sharp cutoff at 20.7KHz, but that's still plenty high enough to sound really good. 112kbps Ogg has a cutoff at 17.9KHz, but that's still better than AAC!

Shayne

!!!!!

The 20.7kHz cutoff is probably where the engineers lowpassed the master recording for CD production! Either that or it's just the highest frequency that you'll find distortion at.

... and now I see it was a joke. lol

[xzxzzx]

Originally posted by ScorLibran
Disclaimer...My apologies of anyone took my last post seriously. It was intended humorously.

For the results of double-blind sound quality tests, check out this 128kbps Listening Test (http://www.hydrogenaudio.org/index.php?showtopic=11936&) conducted late last year.

These four are statistically tied for first place, but...
MPC > AAC > WMA Pro (not normal WMA) > Vorbis Well, this is true, but there are two qualifiers to that: Vorbis may be worse at 128K, but that doesn't mean it's worse at 192 or any other bitrate.

Furthermore, from my lurking over at hydrogenaudio, it seems there is some concensus that Vorbis tends to get lower scores because more people are familiar with the format, and have, in a sense, "trained" their ears to hear vorbis's artifacts/distortions.

[ScorLibran]

You're first point is true indeed. A format that does well at 128 may or may not compare as well with other formats at higher (or lower) bitrates. Another listening test would have to be done at the target bitrate to know.

(Also, keep in mind that those four formats are statistically tied for first place...they are essentially *equal* in sound quality, as their ANOVA error margins overlap.)

I'd tend to agree with your second point as well, though that's pretty hard to measure conclusively. Vorbis has been tested in quite a few tests by many of the same people, so "format-specific artifact awareness" is a viable possibility. Still, since the test results are as close as they are, it's safer to say they tied than "Vorbis is worse".

The best bet to be totally sure is for each person to do their own tests, since what may sound fine to me at 160, for instance might be bad to someone else's ears. But for people with no time to ABX, using test results as at least a "general guide" would be worthwhile.

[xzxzzx]

True, true, true Scor.

[morgado]

Well, first I always used mp3, as most of people did and do ... then I started to come around in Winamp Forums, then I got a light in mey head, I started trying Ogg ... then comes FLAC !!!
FLAC RULEZ !!!!!!

[xzxzzx]

FLAC == pointless, unless you're transcoding for other stuff like an iPod.

[fwgx]

I did use ogg but then got a Nomad Zen tat doesn't like ogg so now I encode MP3's at --alt-preset insane :cool:

[ScorLibran]

Originally posted by xzxzzx
FLAC == pointless, unless you're transcoding for other stuff like an iPod.
...or for archival. That, and the need for many transcodes are my excuse. For general listening, there are definitely better options than FLAC's 800kbps+ in the realm of lossy.

MP3 hits transparency at around 200kbps or up. Vorbis, AAC and MPC do the same at around 160-180kbps. And even 128kbps is still very acceptable for most people, especially for portable players.

[xzxzzx]

Originally posted by ScorLibran
...or for archival. That, and the need for many transcodes are my excuse. For general listening, there are definitely better options than FLAC's 800kbps+ in the realm of lossy. What's the point of perfect archival if all you're going to do is listen to it, and Ogg -q7 is transperant to you?

That having been said, I think I'm going to rerip all my music to FLAC, but I also think I'm going to get an iPod, so I have an excuse.

[ScorLibran]

Bit-perfect archival for secondary backup, plus for an accurate transcoding source. Those reasons are really one reason with two goals, I guess.

I encode to several different (lossy) formats for testing purposes (my own and to help others), so lossless is required. But I'm the exception to the rule. "Normal" people don't need FLAC unless they have mass quantities of HD space they need to fill up. ;)

[xzxzzx]

But if they want to fill HD space, straight WAVs work even better...

Anyway, I stand behind my original statement: Transcoding is the only purpose for lossless compression.

After all, what's the point of backups, besides transcoding?

[Gobd]

Monkeys audio is the best :) Loseless is what everyone should be using now, everyone has plenty of harddrive space.

[xzxzzx]

That's not true. Furthermore, I want 400 CDs on my iPod, damn it.

[ScorLibran]

Among other reasons against it, lossless cannot be streamed over limited bandwidth, hence yet another need for lossy formats.

Monkey's is good, but it has 0 hardware support. My FLACs play in my car just fine. They would on the Rio Karma portable player. as well.

[whiteflip]

just because YOU have a huge hard drive space doesn't mean I do. I could have a 6gb hard drive space and need to encode in the best quality at the lowest bitrate. say 96kbps hahaha.

Which sounds better? 24kbps stereo vorbis or 24kbps stereo aac?

[xzxzzx]

24? I don't think either will encode that low.

[xzxzzx]

Yeah, but either will sound like total crap. Maybe Speex could work ok...

[QuantumKnot]

Originally posted by xzxzzx
Furthermore, from my lurking over at hydrogenaudio, it seems there is some concensus that Vorbis tends to get lower scores because more people are familiar with the format, and have, in a sense, "trained" their ears to hear vorbis's artifacts/distortions.

Definitely. Vorbis has a distinct high frequency boost at quality levels lower than 5 I believe and everyone who is familiar with it will pick it out.

Personally I think Apple should have tried to support many other codecs as well as Vorbis. Ultimately it's up to their marketing department rather than their tech one...AAC has the big word, MPEG-4, behind it while consumers tend to not know who is Frank Klemm or Xiph.Org's Monty. ;)

[ScorLibran]

Hi QuantumKnot.

Originally posted by QuantumKnot
Definitely. Vorbis has a distinct high frequency boost at quality levels lower than 5 I believe and everyone who is familiar with it will pick it out.

Personally I think Apple should have tried to support many other codecs as well as Vorbis. Ultimately it's up to their marketing department rather than their tech one...AAC has the big word, MPEG-4, behind it while consumers tend to not know who is Frank Klemm or Xiph.Org's Monty. ;)
All the more reason Xiph.org should put together a marketing department. I need a break sometimes. ;)

(J/K...I'm a former Vorbis zeolot...not a current one.)

Seriously, for knowing which formats are the best at various bitrates, people should check the Listening Tests (http://www.hydrogenaudio.org/index.php?act=SF&f=40) page at Hydrogenaudio. Look for the threads marked "Finished".

But I'm not aware of any double-blind 24kbps tests, though. To know which format is the best at this bitrate, try ABC/HR (http://ff123.net/abchr/abchr.html). You can do double-blind ABX tests and subjective ratings in one tool.

[QuantumKnot]

Originally posted by ScorLibran
Hi QuantumKnot.

Hi ScorLibran. Nice to see a familiar face here. :D


(J/K...I'm a former Vorbis zeolot...not a current one.)

So are you a Winamp zealot? :D

One other artifact of Vorbis that lots of people seem to notice is stereo separation. Though it can be fairly obvious to the ear at low bitrates (listen to some q -1 or q 0 Vorbis files), it seems there are people who are sensitive to subtle degradations in stereo and they generally listen for it.

I think it is something that Monty is going to fix. For those who don't know who Monty is, look at my previous reply. jk :D

[J_Bloggs]

A OGG stream low enough to not buffer on my 56k modem is listen-to-able. A similer mp3 file hurts my ears/can't even get them to not buffer.

Eh, they all have subtle differences, if you don't like one use the other.

[whiteflip]

reason i ask about the 24kbps is because some people still have the 28.8 connections. Even though all their hardware and provider is setup for 56 their phone lines can only handle 28. Like me.

[xzxzzx]

I still don't think ogg or even Mp3 will encode that low. The lowest I'v ever heard of is 32. I think the only thing that will encode that low is maybe Mp3PRO, but 24 for music wouldn't be listenable. It'd sound like utter shit. You could encode speech that low, it'd be somewhere around telephone quality, I think.

[whiteflip]

The proof is in that which you cherish most.

[ScorLibran]

Ogg Vorbis (v1.0, v1.0.1) at -q -2 is 32kbps nominal, but the encoding specification may support even lower rates.

The specification for MPEG-1 Layer III (MP3) is 32kbps to 320kbps. But I thought I had seen a reference somewhere to a spec saying that 8kbps was the minimum, but I can't find it at the moment.


Edit: These minimums may be lower with sample rates lower than 44.1kHz, one-channel, 8-bitdepth encoding, etc.

[xzxzzx]

Yeah, scor, I think you'r right. Remember people, that's -q (-2), as in two less than 0. Ogg don't do super-low bitrate, it's not designed for it. I could have sworn that Mp3's minimum was 32, but you may be right that lower sample rates and or bitdepths would result in lower than that.

[Stupendous Man]

This is about transparency and what it practically means for the-more-than-the-normal-user
but-less-than-a-die-hard-audiophil-listener:

Originally posted by ScorLibran
A format that does well at 128 may or may not compare as well with other formats at higher (or lower) bitrates.


But as I take it, with all good lossy formats there comes sooner or later a point when transparency is reached. Doesn't that mean that these formats are in terms of their audio quality practically equivalent (if you don't mind of +/- 30 kbps or something)?

Transparency in my understanding means that the music is 'CD-alike' and since all good lossy formats come to this point, albeit with varying bitrates, they are factually EQUAL, not considering slight differences in space. Is it that what you were saying below?

Originally posted by ScorLibran
MP3 hits transparency at around 200kbps or up. Vorbis, AAC and MPC do the same at around 160-180kbps.


PS: I am looking for a new format but superior compatibilty of MP3 seems to get hold of me again before I even try to.

[will]

I only see limited usefullness for anything more than mp3 alt preset standard. To me its totally transparent and everything supports mp3, and everything always will.

[gaekwad2]

2 words:
gapless playback

(And for me Vorbis 1.1rc1 does reach transparency at lower bitrates than Lame.)

[xzxzzx]

Originally posted by will
I only see limited usefullness for anything more than mp3 alt preset standard. To me its totally transparent and everything supports mp3, and everything always will. This kind of reasoning is why we still all type on QWERTY keyboards.

[will]

gaekwad2: three words: Directsound output config

xzxzzx: And what is wrong with qwerty? The layout of keys is an arbitary thing. Any preference to other layouts is purely personal. Its as good a standard as any.

[gaekwad2]

Originally posted by will
gaekwad2: three words: Directsound output config
Still produces a small but audible click at track changes.

[ScorLibran]

Originally posted by Stupendous Man
But as I take it, with all good lossy formats there comes sooner or later a point when transparency is reached. Doesn't that mean that these formats are in terms of their audio quality practically equivalent (if you don't mind of +/- 30 kbps or something)?
Yes, you're correct. Transparency is transparecy. Anything above transparency threshold (with the exception of a few "problem samples") is equal across all formats that can attain transparency.

Where that threshold lies is important for lossy audio encoders and their developers. And because of that, early next year I'm going to conduct a large-scale listening test (using ABX and the ITU-R BS.1116-1 testing standard) to determine the transparency threshold for all major lossy formats (MP3, AAC/MP4, Ogg Vorbis, WMA and MPC).

Originally posted by Stupendous Man
Transparency in my understanding means that the music is 'CD-alike' and since all good lossy formats come to this point, albeit with varying bitrates, they are factually EQUAL, not considering slight differences in space. Is it that what you were saying below?
Yes. After reaching transparency threshold, it's a matter of efficiency, measured as sound quality divided by bitrate. Bitrate directly affects filesize.

Originally posted by Stupendous Man
PS: I am looking for a new format but superior compatibilty of MP3 seems to get hold of me again before I even try to.
That's indeed the best reason to stay with MP3. And if it weren't for a need for gapless playback on all platforms, I'd be using MP3 myself.

Originally posted by gaekwad2
Still produces a small but audible click at track changes.
I posted in another thread that the mpg123 MPEG decoder (http://esc17.midphase.com/~calmerc/chris/audio_encoding/in_mpg123_118ot55.zip) provides the only true audibly gapless playback for MP3 in Winamp (that I've confirmed myself, anyway). Many people may be able to tweak Directsound to get good results, but more success can be found with the correct input plugin, rather than output plugin. Gaps are not a playback issue, but an issue of decoding (actually an issue of encoding, but they can be addressed at decode time).

[Starbucks]

I love lossless :)

[xzxzzx]

Originally posted by will
xzxzzx: And what is wrong with qwerty? The layout of keys is an arbitary thing. Any preference to other layouts is purely personal. Its as good a standard as any.

Because qwerty forces your hand to move significantly more than a proper layout would. In an optimal layout, the most used keys would be on the home row, keys would be positioned in such a way that - when typing typical words - which hand is alternated.

Dvorak (not named for the order of keys, but after the inventor) is provably better than qwerty.

See http://www.mwbrooks.com/dvorak/

[Stupendous Man]

Thanks for your thorough answer. :up:


Originally posted by ScorLibran
Yes, you're correct. Transparency is transparecy. Anything above transparency threshold (with the exception of a few "problem samples") is equal across all formats that can attain transparency.

That about clears it up. So Mp3 is in terms of its audio quality as good as any lossy format (and also any lossless format strictly spoken) if only the point of transparency is reached. At that point everything has become practically equal. That's what I wanted to know.

And with reaching the point of transparency a technology has got to its natural end, natural meaning that the real obstacle for any improvement doesnt exist any more in technological but now human restrictions. The addressee for further improvement simply isn't there anymore, advancing the technology would become an end in itself.

I noticed the same development in cellular phones as their size hasn'
t shrunk anymore for some time - they simply can't get much smaller than the size of the hand nor does the sharpness of the human eyes allow the display to get any smaller.



Originally posted by ScorLibran
Yes. After reaching transparency threshold, it's a matter of efficiency, measured as sound quality divided by bitrate. Bitrate directly affects filesize.

Personally, I can imagine that all this business of choosing different lossy formats will come to an end in the next few years.

HD space is getting bigger and bigger at an affordable price, making compressing files finally obsolete. The whole problem will be solved not by further refining the encoding technologies but by enhancing the disc space at one's disposal. Kind of tackling the problem by the supply side.

'Only' at Internet exchange platforms lossy formats will still be widespread and popular for a long time.

Just my guess.

[k_rock923]

Wasn't the qwerty from the typewriter and wasn't it designed so that people didn't type too fast on a typewriter and jam up everything?

[Rellik]

QWERTY doesn't necessarily slow you down, it was designed to allow for most words (especially commonly paired letters) to be split up between the left side and the right side, thus preventing most jamming.

more info (http://en.wikipedia.org/wiki/Qwerty)

[xzxzzx]

I don't know who wrote that part of the article, Rellik, but it is generally accepted that you can write faster with Dvorak keyboards compared with qwerty.

Qwerty was not, in fact, designed to split up the letters on the actual keyboard - it was designed to split up the letters in the punching mechanism of the typewriter.

Note that I've spent about 5 hours practicing with a Dvorak layout, and I'm already seeing how much nicer it will be once I switch over completely.

[gaekwad2]

Originally posted by Stupendous Man

That about clears it up. So Mp3 is in terms of its audio quality as good as any lossy format (and also any lossless format strictly spoken) if only the point of transparency is reached. At that point everything has become practically equal. That's what I wanted to know.
That depends on whether for you the codec reaches transparency at all.
I.e. WMA standard doesn't for me no matter what bitrate I choose, and for some (very few) people mp3 doesn't either.

[Russ]

Originally posted by xzxzzx
I don't know who wrote that part of the article, Rellik, but it is generally accepted that you can write faster with Dvorak keyboards compared with qwerty.

It's not generally accepted, mainly because you can't scientifically prove it's quicker (I personally don't think it is). As the good article says, the main advantage is that it reduces movement, and hence is more comfortable, plus the learning speed is considerably less. It's all explained very well in P. Buzing (http://www.cs.vu.nl/~pcbuzing/Articles/keyboards.pdf), which is referenced from the Wikipedia article.

[xzxzzx]

Originally posted by Russ
It's not generally accepted, mainly because you can't scientifically prove it's quicker (I personally don't think it is). As the good article says, the main advantage is that it reduces movement, and hence is more comfortable, plus the learning speed is considerably less. It's all explained very well in P. Buzing (http://www.cs.vu.nl/~pcbuzing/Articles/keyboards.pdf), which is referenced from the Wikipedia article. It may be explained well, but it is based on questionable data at best. For instance:
A The loads on the right and left hands are equalised
B The load on the home row is maximised
C The frequency of alternating hand sequences is maximised and the frequency of same-finger
typing is minimised.
The DSK fulfills the first two constraints, but the QWERTY layout is better at constraint C.
This seems to be lifted directly from "The Fable of the Keys", which has plenty of refutations available (http://www.mwbrooks.com/dvorak/dissent.html).

For an only somewhat related, but interesting page, take a look at:

http://www.visi.com/~pmk/evolved.html

[whiteflip]

Originally posted by xzxzzx Note that I've spent about 5 hours practicing with a Dvorak layout, and I'm already seeing how much nicer it will be once I switch over completely. [/B]

Let me know when you hit 300 words per minute.

Originally posted by gaekwad2
That depends on whether for you the codec reaches transparency at all.
I.e. WMA standard doesn't for me no matter what bitrate I choose, and for some (very few) people mp3 doesn't either.

Yeah but you have to consider the rate at which transparency is reached and the ammount of processor power it takes for decoding and such. Even though Mp3 does take up more space to get transparent it takes far less processor power than other codecs and thats important for portable players and such.

[saivert]

What about SBR (Spectral Band Replication) used in MP3pro and AACplus? That may give people a good deal with low bitrates like in the 32kbps-64kbps range.

[gaekwad2]

Mp3pro is dead, AACplus/HE-AAC may be useful if it ever gets supported.

Otoh it's only used up to 96kbps, and according to one of Ahead's HE-AAC devs Vorbis 1.1 is currently superior at 96.

Imo it would only be useful for cheap small flash based players or mobile phones.

[mark]

hit 300 yet xzxzzx? also, how hard is it to type xzxzzx on dvorak? :)

[whiteflip]

http://systems.webopedia.com/FIG/DVORAK.gif
harder than on qwerty

[Kinetik*]

I think it's safe to say that every platform should just go with OGG by default

[nybergh]

Originally posted by Kinetik*
I think it's safe to say that every platform should just go with OGG by default Does the xbox version of HALO use Ogg as well as the PC version?

[xzxzzx]

Originally posted by whiteflip
[Image] (http://systems.webopedia.com/FIG/DVORAK.gif)
harder than on qwerty Actually, it's easier due to the alternation of hands, though it's harder with one hand.

[whiteflip]

Originally posted by xzxzzx
Actually, it's easier due to the alternation of hands, though it's harder with one hand.

I was taking into consideration the fact that it took me 2 minutes to find the letter Z :)

Not every platform should play OGG because its so processor intensive. Tiny Flash players for example would have shitty battery life and require more expensive hardware. But everything with a decent chip (new consoles, PC's, maybe PDA's) should use OGG for audio because of the quality and the freeness. I don't understand why companies don't enjoy the freeness.

[gaekwad2]

Because if they pay for a license they feel safer from patent claims.

Plus many corporates think free software is unprofessional, they still view all open source projects as nerdy hobby stuff.

[ScorLibran]

Don't rely on companies to "provide" Vorbis support. Some open-sourcer always comes through as a hero.

With the free MCA module, for instance, any Palm 5 OS device (including my Sony Clie NX80) can play Vorbis tracks up to at least -q 4, plenty above the transparency threshold. (Using Aeroplayer, or my preference, Pocket Tunes.)

Vorbis wasn't written or managed to be a licensed corporate product, just ask Emmett Plant (former CEO of Xiph). It's meant to be a top-shelf audio encoding format, and that's it. Whoever can make use of it, more power to them.

The world's strongest anti-Vorbis forces at Hydrogenaudio pushed last year to topple Xiph's claim that Vorbis is patent-free. But instead of accomplishing their goal, they established more firmly that Vorbis is indeed free of any patent infringement...much more sure about it than anyone was going in to that fight, anyway.