Which lossy format are you using? Options

[Benjamin Lebsanf...]

vqf better than mp3 or vorbis and mpc ? What speakers are you using :eek: ???

[Negative Zero]

I mostly stick to LAME --alt-preset standard -Y encoded MP3's for most of my digital audio files because of the size, quality, and hardware playback support (on my iRiver SlimX). Every once in a while, I do encode a few tracks into MPC, but until I see hardware support for that format, I can't be bothered to give up MP3 in favor of MPC.

[lucpes]

Originally posted by Dezibel

...a few albums sounds "better" than the original cd.

And here we go to CD+ quality

[Dezibel]

Originally posted by lucpes


And here we go to CD+ quality

i forgot the smiley

Dezibel

[Dezibel]

Originally posted by silver_cpu
Dezibel, I think that the settings for mpc that you settled on are overkill.  You might want to try out --xtreme.  This setting was designed for people who were not quite totally satisfied with --standard.  It is the setting that I use, and I'm totally happy with it.  Basically, it's one step above --standard, but changes some of the presets, such as the ltq (or ath, whichever you prefer to call it) to give you slightly higher quality, and comparable bitrates.  Try --xtreme some time, or one of the new number quality levels, and just keep going up until you figure out what's best for you.  Note, however, that even though you can go up to 10, 6-7 is more than enough, and anything above 8 (and in most cases less) is truely overkill, and a waste of hdd space.

i agree that --xtreme sounds transparent from point of frequency spectrum on most songs. i did some ABX tests a few weeks ago where mpc decoded back to wave sounds almost identical to the source wave at bitrates higher than --standard. but for me on hq headphones [akg k141 or 240df] the sound of an mpc compared with the wave file sounds a little bit "thin". i need more compression or pressure to became the same "feeling". and i'm sure that it is not an psychical problem. --insane or --braindead [--quality 7 or 8] sounds good for me from point of "pressure".

...you can compare with image compression. while jpeg compression is designed to save subjective image quality. the jpeg filters on photoshop looks like the colors where displaced a little bit. and the sharpness is lost. the jpeg filters on my linux system [using gimp or the convert command on commandline] looks like the colors are fine, sharpness are fine, but there are little artifacts at fonts and little details. you can life with 90 percent jpeg quality or save the last 2 bits using an lossless compression like png. if ypu want an preview pic on your website for fast download than you have to go with small files. but if you want to archive anything you have to decide how many bits you can spend to save hard work for generations

Dezibel

[God eat God]

Aim bit-rate: ~150 - ~170

Type of music: very complex : allot of black-metal (Dimmu Borgir), metal (Grave Digger, Judas Priest, Slayer, Metallica)

Sound quality: best for the bit-rate. i don't need precise average bit-rate. i used mp3 LAME --alt-preset standard -Y, but it ended ~175kbs (a bit too high, but it's not THAT important), and more important, a bit 'thin'. it didn't sound the same.

Sharing / portability: none. The only person that will hear what I've encoded is myself, on my PC, with my excellent headphones. so it's ok to use mpc, even dough it's not popular/can't use it on a portable player.

So, will mpc --standard will give me better quality for the specified music (it's ok if it ends with SLIGHTLY higher bit-rate...)? Will OGG -q4.99? Any other format?

[Dibrom]

Originally posted by God eat God

So, will mpc --standard will give me better quality for the specified music (it's ok if it ends with SLIGHTLY higher bit-rate...)?

Yes.

[b]

Will OGG -q4.99? Any other format?

I'm not sure. I haven't done any testing with the later builds of rc4 or 1.0. There should be a q val that will give you similar quality, but I'm not sure sure what it'd be right now.

[Dibrom]

Originally posted by Dezibel
i agree that --xtreme sounds transparent from point of frequency spectrum on most songs. i did some ABX tests a few weeks ago where mpc decoded back to wave sounds almost identical to the source wave at bitrates higher than --standard. but for me on hq headphones [akg k141 or 240df] the sound of an mpc compared with the wave file sounds a little bit "thin". i need more compression or pressure to became the same "feeling". and i'm sure that it is not an psychical problem. --insane or --braindead [--quality 7 or 8] sounds good for me from point of "pressure".

Umm... am I understanding this right? You did abx tests finding that the encoded output was perceptually identical to you, yet you are still implying there is some difference that you can hear, but that you can't abx (so you basically can't hear it)?

Heh.

It's always the same. "Thin", "feeling", "lack of bass", etc. Yet no positive abx results to back any of it up. I guess some people will just never give up their delusions even after proving to themselves in a blind test that they can't hear the difference.

[God eat God]

Originally posted by Dibrom


Yes.

[b]

I'm not sure.  I haven't done any testing with the later builds of rc4 or 1.0.  There should be a q val that will give you similar quality, but I'm not sure sure what it'd be right now.

10x dibrom. I thought I heard somewhere that mpc isn't optimized for metal... anyway, so this is the best codec for < 170kbs black-metal / metal out there? (What about AAC? or anything else?)

what about the new OGG 1.0 and metal, anyone?

[Dibrom]

Originally posted by God eat God


10x dibrom. I thought I heard somewhere that mpc isn't optimized for metal...

Dunno where you heard that. If you did actually hear it, I'm pretty certain that it was just some unverified rumours (as usual).

[b]anyway, so this is the best codec for < 170kbs black-metal / metal out there?

Well, mpc -standard is generally transparent on most files to most people (yes, even the "golden ears"), at a bitrate around 160kbps. Take that as you want, but that means that it's certainly good at metal, yes.

[b](What about AAC? or anything else?)

AAC is also a good choice if you use a good encoder. PsyTEL AAC is usually the encoder of choice since it is the highest quality "publically" available AAC encoder.

[b]what about the new OGG 1.0 and metal, anyone?

Vorbis should do well also.

Again, MPC has kind of a proven track record with the -standard setting like I said, but you shouldn't do bad with AAC or Vorbis either.

The quality of all of these codecs are usually so high that the decision on which format to use falls to the other aspects/features of the encoder that one will use.

[Neo Neko]

Originally posted by lucpes


And here we go to CD+ quality

What you mean that if WMA is CD quality at 64Kbps and I raise the bitrate I get no increase in quality? Such blasphemy?! Lord Bill shall hear of this. We will smite you all. Smite smite smite. Love that word. Blame the new Microsoft math ™. Just like the old math only buggier and prone to incorrect output. LOL

OGG for me with a little MPC/AAC on the side.

[SometimesWarrior]

Originally posted by Dibrom
Umm... am I understanding this right?  You did abx tests finding that the encoded output was perceptually identical to you, yet you are still implying there is some difference that you can hear, but that you can't abx (so you basically can't hear it)?

Heh.

It's always the same.  "Thin", "feeling", "lack of bass", etc.  Yet no positive abx results to back any of it up.  I guess some people will just never give up their delusions even after proving to themselves in a blind test that they can't hear the difference.

Well I do remember this post, where Dezibel actually made test samples and ABX results available. Was anything ever done to follow up on this report?

[KikeG]

I use MP3, LAME --alt-preset standard --lowpass 18 is good enough for me. I guess --lowpass 17 or -Y could be too, I would have to test it seriously.

However, when doing "quality" listening, sometimes I still prefer uncompressed, it gives me a "mental" feeling of greater "quality", or that I feel "safer" about the quality.

But I guess it's just a psychological thing, since I can't say it is better on a objective blind test, and I really believe that for me they are undistinguishable on regular music.

[Dibrom]

Originally posted by SometimesWarrior
Well I do remember this post, where Dezibel actually made test samples and ABX results available. Was anything ever done to follow up on this report?

The problem is that the abx test results were not very significant. 8/10 trials in only 1 test isn't really enough to be "sure" I don't think. Secondly, one of the trials where he got 7/10 on the first try, he later was only able to get 8/15.

The results just aren't really strongly enough in favor of an actual difference being heard reliably. One way to increase the significance would be to increase the number of trials, or to repeat the tests multiple tries. None of that has been done though..

[stoff]

Ack... pick your poison, right?

Well, I'm quite happy with LAME --alt-preset standard for my Rio Volt. (Thanks a lot to Dibrom!)

I don't listen to music on my computer so all good albums that I come by are just copied the old fashion CDR-way. Hardware support dictates everything for me, so I'm looking forward to ogg (or even better MPC support) on a PocketPC with 512MB flash ram!

Regards, stoff

[wtkwest]

Originally posted by silver_cpu
If you simply must have minute files, then choose ogg or wma, maybe mp3pro if it turns out to have appeal to the masses in the end. 

I wasn't after particularly small files. What I was after was to basically prove to myself that mp3 @ cbr 192 was really a decent format, and that ogg would be the next likely candidate...

Trust me, my results surprised me as well...

For my next test, I will try mpc and aac, and start with a CD and then a tape captured at 44.1kHz... I have a friend who is an avid ogg user who suggested that ogg may not work as well with 48kHz files... But that as a complete guess...


My long term goal is to get a halfway decent capture card (again my sound card ain't so great), and eventually move stuff to mp3.com... So capturing sound is pretty high on my list... ...and mp3 will probably be my end distribution format... ...and I will wonder "What if..."

Thanks for the suggestions...

[Dezibel]

Originally posted by Dibrom


The problem is that the abx test results were not very significant. 8/10 trials in only 1 test isn't really enough to be "sure" I don't think.  Secondly, one of the trials where he got 7/10 on the first try, he later was only able to get 8/15.

The results just aren't really strongly enough in favor of an actual difference being heard reliably.  One way to increase the significance would be to increase the number of trials, or to repeat the tests multiple tries.  None of that has been done though..

ufff,

forget my abx tests. forget anything i've said about mpc quality.


mpc sounds for me lifeless. andrees encoder, franks encoder, version 1.06, standard, xtreme and braindead minSMR3. everything sounds lifeless for me. i can't significantly abx what i miss. so my experiences are no reason to retune the encoder.

Dezibel

--
plain old audio cd for me please!

[CiTay]

Lifeless? This is a completely psychological phenomenon, IMO.

[JohnV]

Originally posted by CiTay
Lifeless? This is a completely psychological phenomenon, IMO.

And if it _sounds_ lifeless, it should be possible to ABX. In principle any difference whether "lifeless" or what ever should be ABXable.

Another blind test you can try is the "odd man out" -test or ABA. Try playing 2 originals and one encoded (decoded wav) in Winamp. Shuffle between tries and try to pick the encoded 5-6 times in a row. You can let the music flow and tracks repeat and just concentrade on music, and check when in your opinion the track sounds "lifeless".

[Phobos]

omg who voted fotr wma and vqf???

there should be a sticky about:

64kb/s in no way is CD quality!!! cuz this codecs suck even more at high bitrates...

Glad to say crapiest mp3PRO still has no votes...

[Dezibel]

Originally posted by CiTay
Lifeless? This is a completely psychological phenomenon, IMO.

...just for fun:

encode your favorite albums lossless [flac...] and also mpc [with your favorite switch --standard, --xtreme...]. hear 2 or 3 days nothing but these lossless albums from your harddisc using your favorite headphone and your stereo.

on 3th day after many many ours hearing these albums [lossless] start one of these albums in lossy quality. then you will know what a psychological phenomenon is and what not.

...you can acclimate everything.

Dezibel

[fewtch]

Dezibel... I think you make a good point regarding the main weakness of a program like PC-ABX: It only takes a short time factor into account. I've always felt this was a weakness (of the usual ABX methods) too, but there's always somebody that produces various arguments knocking the theory. Which reveals yet another maxim: Anything can be successfully argued too! A simple debate class will teach that...

[CiTay]

I'm still sticking to what i said.

Did you ever hear of something called the "placebo effect"? When you start to play the lossless files in that test of yours, maybe you're unconsciously expecting a different sound from them.

This can have a bigger influence than you may think. In 1998, there was a test performed on more than 2000 people, with the antidepressant "Prozac" and a placebo. You can read about it here: http://journals.apa.org/prevention/volume1/pre0010002a.html

The outcome was that half of the drug's effectiveness is on account of the placebo, and only one quarter of the drug response is due to the active medication.

Maybe now you understand why we rely on ABX results, not on feelings.

[Phobos]

Is placebo effect just an effect i can get rid off or is it permanent more like a disease???

I started to try vorbis 1.0 against psytel AAC and switching from one to another in ABX tests along with my mediocre ears i got to a point i dont know wich one gets closer to the original wav. man i cant even trust my brain or ears, imagine how anoying it is...

[CiTay]

Originally posted by Phobos
Is placebo effect just an effect i can get rid off or is it permanent more like a disease???

No, it's not a disease... it's an effect, as the name implies. I'll try to explain it differently: A little kid crashes with his bike and now his knee hurts. When he comes home, his mom gives him a teaspoon with medicine and promises that it will ease the pain. The little boy immediately feels better. But all his mother gave him was water with a little sugar in it. No active ingredient. It was only in his mind.

This does really happen, to adults as well. Due to the huge effect this "wishful thinking" can have on humans, it is widely used in medical treatment.