[lindrone]

Quote:

Originally Posted by hackeron No. Even if you set the max possible bitrate the encoder allows, it will cut off inaudiable frequencies straight away -- thats nothing to do with the bitrate. -- ogg is more agressive in this respect (atleast early versions used to be).

No... different encoders has different spectrum analyzers that determine what frequency is really inaudible, and what effects of frequency overlap needs to be retained or not.

There's a reason why LAME encoder makes better MP3 files at the same bitrate as some other encoder... and there are some truly crappy encoders out there that's just fast & dirty. One such example is Xing, who used to make AudioGrabber or something... their own encoder was super fast, and super crappy at determine what needs to be kept and what gets cut off.

So even though all encoders will cut off what they determine to be "inaudible"... each encoder's spectrum analyzer will have a differing opinion on what constitutes "inaudible".

 

[hackeron]

Quote:

Originally Posted by lindrone No... different encoders has different spectrum analyzers that determine what frequency is really inaudible, and what effects of frequency overlap needs to be retained or not.

There's a difference between inaudiable frequencies and over lapping frequencies. The initial rule is to remove certain frequencies no matter what bitrate is set.

Quote:

There's a reason why LAME encoder makes better MP3 files at the same bitrate as some other encoder... and there are some truly crappy encoders out there that's just fast & dirty. One such example is Xing, who used to make AudioGrabber or something... their own encoder was super fast, and super crappy at determine what needs to be kept and what gets cut off.

Hmm, really? -- I havent seen any frontends that have their own backend. Why would anyone waste months (atleast) to make yet another mp3 encoder is beyond me... -- I guess this would be the only exception. Making an encoder isnt exactly easy.

Also, even a 166mhz machines can encode with lame in realtime -- is speed still a concern?

Quote:

So even though all encoders will cut off what they determine to be "inaudible"... each encoder's spectrum analyzer will have a differing opinion on what constitutes "inaudible".

Yes, they do. What I'm talking about is a general starting rule. Every encoder has a set of frequencies they cut off no matter what bitrate you set. The ones that are definitely inaudiable by the human ear. Rest are calculated with complicated algorithms to find out overlaps, etc as you said.

What I'm talking about is this initial rule that is similar on all lossy compressors, some are more agressive than others, but the initial rule is generally very, very similar. From there the algorithms kick in to eliminate what they *think* the person won't be able to hear, and thats where most of the data is lost.

If I'm not wrong, lame eliminates more mids than highs and lows. Makes sense why it sounds worse than ogg on the shures. Wma on the other hand colours the sound and removes even more mids -- so even 64kbps wma sounded tolerable on the EX71 that simply dont have clarity in the mids.

 

[lindrone]

Quote:

Originally Posted by hackeron Hmm, really? -- I havent seen any frontends that have their own backend. Why would anyone waste months (atleast) to make yet another mp3 encoder is beyond me... -- I guess this would be the only exception. Making an encoder isnt exactly easy.

AudioGrabber is a pretty old product, don't even know if they sell it anymore. It was made a couple years back when it was easier to advertise "speed" of the MP3 encoders because a lot of people had slower computers.


Quote:

Also, even a 166mhz machines can encode with lame in realtime -- is speed still a concern?

What do you consider "real time"?.. You mean encoding as fast as it plays back? Or something else? LAME is significantly slower than encoder from Fraus... can't spell it.. anyway, the German company that holds the patent and supplies most of the encoders for software out there. LAME is probably by far the slowest encoder out there right now.


Quote:

Yes, they do. What I'm talking about is a general starting rule. Every encoder has a set of frequencies they cut off no matter what bitrate you set. The ones that are definitely inaudiable by the human ear. Rest are calculated with complicated algorithms to find out overlaps, etc as you said.

Here is another tangent discussion on what is really "audible by human ear"... you can have that discussion elsewhere in the another forum.. this thread just isn't a very good place for it. However what all of those lossy codec determine as "inaudible" isn't adequate anyway. Hence a lot of us can still tell the difference between lossless and lossy codec.

Try something lossless on your E5's.. they'll sound better.

Either way, the point was that different encoders will have different results at the same bitrate, so what encoder you use does matter in matter of quality. Bitrate isn't the only determining factor, that's all.

 

[hackeron]

Quote:

Originally Posted by lindrone What do you consider "real time"?.. You mean encoding as fast as it plays back?

Yes.

Quote:

Here is another tangent discussion on what is really "audible by human ear"... you can have that discussion elsewhere in the another forum.. this thread just isn't a very good place for it. However what all of those lossy codec determine as "inaudible" isn't adequate anyway. Hence a lot of us can still tell the difference between lossless and lossy codec.

What I'm talking about is scientifically what frequencies the human ear cannot hear -- *NOT* what detail loss is noticeable or not.

Can you hear 23khz? no!
Do these frequencies exist on an uncompressed source? yes!

Therefore they get removed by default. Rest is pretty much removing detail from certain frequencies. This was a response to Mr Iriver about encoders saving most space by removing extensions. They dont -- they remove detail from frequencies. Extensions get removed initially no matter what encoder or setting.

Now what detail loss is tolerable or not is a fragile subject when comparing encoders and bitrates and this thread is certainly not the right place. However Mr Iriver said he can tolerate 64kbps for some music on the E3c -- you dont see anything wrong with that?

Quote:

Try something lossless on your E5's.. they'll sound better.

I did. Sounds as enjoyable as 256kbps MP3, although I can hear a difference to uncompressed, just cant quite put my finger on it so good enough. -- My source might not be good enough though seeing as many people say 320kbps sounds bad.

Quote:

Either way, the point was that different encoders will have different results at the same bitrate, so what encoder you use does matter in matter of quality. Bitrate isn't the only determining factor, that's all.

Encoders? yes. Programs? no. For mp3 there is lame and for ogg there is vorbis (oggenc). Both give very different results, ofcourse. But appart from lame what other mp3 encoders are out there (except for maybe ancient encoders pre lame)?

There were 1 or 2 commercial mp3 encoders, and I have some mp3s from them, from many years back. -- if they are worse than lame, then maybe marginally, But still, who is using them today?

 

[lindrone]

Quote:

Originally Posted by hackeron What I'm talking about is scientifically what frequencies the human ear cannot hear -- *NOT* what detail loss is noticeable or not. Can you hear 23khz? no! Do these frequencies exist on an uncompressed source? yes!

Actually, that's just an average measurement of expected hearing range for the human ear. There's individual variance of what one can hear, some people can hear more than others... although the degree of variance is not large, it is there.

There are also frequencies that has a tangible effect on the sound but you can't hear... it's still there.


Quote:

However Mr Iriver said he can tolerate 64kbps for some music on the E3c -- you dont see anything wrong with that?

Nope, not what I'm saying at all, I'm just pointing out some of what Mr Iriver was talking about isn't the same as you think he was talking about. He started talking about different encoders, not just about bitrates.



Quote:

I did. Sounds as enjoyable as 256kbps MP3, although I can hear a difference to uncompressed, just cant quite put my finger on it so good enough. -- My source might not be good enough though seeing as many people say 320kbps sounds bad.

The "quite can't put my finger on it" difference is a combination result of some lost detail, and what I believe, some of those intangible frequencies that you're not supposed to hear. This is pretty much what differs lossless and lossy compression... also an addictive reason for some of us, although not all of us, to abandone lossy compression forever.


Quote:

Encoders? yes. Programs? no. For mp3 there is lame and for ogg there is vorbis (oggenc). Both give very different results, ofcourse. But appart from lame what other mp3 encoders are out there (except for maybe ancient encoders pre lame)?

LAME is very rarely used outside of the techie/audiophile community. Even though it yields higher quality, most consumers can't tell the difference from one or another. First of all they don't really care, second is that they don't have good enough of equipments to show the difference. All they care about is "how fast can I rip?"

There are very few programs uses LAME as its default encoder. Some programs you can swap out the encoder to use LAME instead. Most software out there uses Fraunhofer's encoder, which was either the patent holder for MP3.. or directly worked for the patent holder (I think the patent holder was actually Thomson or something... I don't know all the legal histories behind it). The list of software would include, at least to my knowledge, iTunes, RealOne, MusicMatch Jukebox, I think even Winamp uses it. Right there you have most of the consumer MP3 software base covered.

Also, I think Frauhofer's encoder is the only one that offers MP3 Pro encoding right now. Means nothing to us... of course.

The only one that I know of which uses LAME as its default encoder, is JRiver MediaCenter, which is a software that's more aimed towards high-end audio (heck, if offers Monkey's Audio support). dbPowerAmp also offers LAME... but otherwise I can't think of a commercial package that has LAME right off the bat.

 

[dura]

Quote:

Can you hear 23khz? no!

Indeed you can't...if you generate a isolated 23 Khz you very likely won't hear it.
But will you hear the difference between a source with information up to say 100kHz copared to one where everything is filtered away at 20 Khz?
According to many sources, yes!
F.i. Marantz and Pioneer did these kinds of experiments over a decade ago, to increase their understanding of how we hear. There where lots of other investigations too (often quoted in theoretical articles if f.i. HifiNews and RR) and reviewers claim to clearly hear the differnces when adding a supertweeter, even if the thing only works above the freq.threshold.

What is a high frequency anyway?
Detail, that's what it is, in reality; A musical instrument produces differences in airthickness.
the isolated sinusoid frequences shown in graphs are nothing but the mathematical results of fourier-analysing music, they do not in reality exist in that form, nice regulair wavelengths, independent of each other.

F.i do a fourieranalyse on the sound of a contrabass and you will find tones up to 35 Khz, co-defining the character of the instrument in a increasingly subtle way.
One simple and likely possible explanation is interference; high-frequencies interact with all kind of other tones with audible results.

 

[lindrone]

Bravo Dura! Great information..!! I didn't even know about that before...

 

[dura]

Thank you Lindrone.
I have the idea there is a lot of misunderstanding about high frequencies on the internet, usually of the "I'm-no-bat" type.

 

[lindrone]

Quote:

Originally Posted by dura Thank you Lindrone. I have the idea there is a lot of misunderstanding about high frequencies on the internet, usually of the "I'm-no-bat" type.

LOL... yes, the "I'm-no-bat" type is very rampant

I always knew that frequencies out of our normal hearing ranges does affect sound, just never really had a clear explanation on how or why. This short explanation really helps clear a lot of things up... Also points towards the fact that there's been pre-existing research on the subject

 

[hackeron]

Quote:

Originally Posted by dura Indeed you can't...if you generate a isolated 23 Khz you very likely won't hear it. But will you hear the difference between a source with information up to say 100kHz copared to one where everything is filtered away at 20 Khz? According to many sources, yes! F.i. Marantz and Pioneer did these kinds of experiments over a decade ago, to increase their understanding of how we hear. There where lots of other investigations too (often quoted in theoretical articles if f.i. HifiNews and RR) and reviewers claim to clearly hear the differnces when adding a supertweeter, even if the thing only works above the freq.threshold. What is a high frequency anyway? Detail, that's what it is, in reality; A musical instrument produces differences in airthickness. the isolated sinusoid frequences shown in graphs are nothing but the mathematical results of fourier-analysing music, they do not in reality exist in that form, nice regulair wavelengths, independent of each other. F.i do a fourieranalyse on the sound of a contrabass and you will find tones up to 35 Khz, co-defining the character of the instrument in a increasingly subtle way. One simple and likely possible explanation is interference; high-frequencies interact with all kind of other tones with audible results.

Hmm, makes sense. But lets look at it logically now.

The frequencies are there on an uncompressed source and they affect other frequencies. Lossy encoders remove them, and since most equipment cant play more than 20khz anyway, all that happens is all negative side effects disappear.

Makes sense actally. I seen some blind tests where a 320kbps MP3 is rated higher than source. If the equipment is asked to play a frequency it cant, it just makes sense it will cause some sort of distortion that is eliminated with lossy compression. Booya!

Very interesting indeed

 

[hackeron]

I have a test for you actually, crank up some music, then get someone to blow a dog whistle. If you think that improves anything, get a super tweater and let it just constantly play something in the background

EDIT: I dont doubt that some can feel very low and very high frequencies, especially since they affect other frequencies. I also read about tests where a super sub woofer was set to play a frequency as low as 10hz --- while no-one could hear it, most felt sick after being in its presence for several minutes.

Not exactly an improvement.

 

[Pov]

Just to get back on topic here, I use the E3C's with no equalization whatsoever and I am very satisfied. OK, once in a while I'll turn on the "bass booster" on my iPod, but then I turn it off again

 

[hackeron]

Quote:

Originally Posted by Pov Just to get back on topic here, I use the E3C's with no equalization whatsoever and I am very satisfied. OK, once in a while I'll turn on the "bass booster" on my iPod, but then I turn it off again

After growing to love the EX71 in the years I had the, I couldnt listen to something completely the opposite without equalization personally

Still, would be interesting to hear what headphones you used before the E3c. -- Good idea on getting back to the subject btw.

 

[Pov]

Quote:

Originally Posted by hackeron After growing to love the EX71 in the years I had the, I couldnt listen to something completely the opposite without equalization personally Still, would be interesting to hear what headphones you used before the E3c. -- Good idea on getting back to the subject btw.

Nothing too spectacular - Senn MX500, Koss Portapros, Sony D66 Eggos, Sony MDR V6. E3C blows them all away, IMHO, though I still do enjoy the Portapros when isolation is not an issue. FWIW, I find the 2 Sony cans to be somewhat shrill and the Eggos to be lacking in bass.

 

[Mr Iriver]

S_Dedalus, Pov, Lindrone, , Hackeron , Dura and the others thanks for all the useful info on compression, and frequencies. By the way when I said I tolerate 64K in some music I did not mean it sounds good. In fact it sounds bad. However if like with the kazaa downloads I can't improve the sound quality on those files with low bitrates, If the song is good, the distortion and the negative artifacts are not going to stop me from enjoying the song. For this reason I rip about all of my music is at 256k mp3 and 160 variable ogg .My flash player does not have enough memory for Lossless tough, and does not play lossless neither.