IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
320 kbps mp3 peaking at 22khz, difference?
xdesirex22
post Sep 10 2012, 02:49
Post #1





Group: Members
Posts: 25
Joined: 20-April 12
Member No.: 99005



I have always used LAME to encode my 320 kbps MP3s, but I know there are other encoders out there that encode 320s that peak at 22khz. Does this make for an increase in quality or audible difference? I have always thought LAME to be the best option, but I have heard from different people that 320s that peak at 22khz sound better on a large party/concert sound system.

also, what are some of the other lossy encoders out there besides LAME?
Go to the top of the page
+Quote Post
Tahnru
post Sep 10 2012, 03:16
Post #2





Group: Members
Posts: 325
Joined: 17-October 05
From: United States
Member No.: 25178



If I may make a recommendation, consider your questions from the standpoint of transparency. Lame is generally accepted to achieve this at -V2.

Further thoughts:


Spend some time poking around the rest of the HA wiki. Be extremely wary of listening to random people about audio matters, especially as it relates to often misunderstood topics like perceptual encoding.
Go to the top of the page
+Quote Post
xdesirex22
post Sep 10 2012, 04:10
Post #3





Group: Members
Posts: 25
Joined: 20-April 12
Member No.: 99005



Not really a fan of the term transparency. First off, it is completely subjective. Different files sound differently to different people. Additionally, it is also VERY dependent on what speakers you are playing your music on and how you have connected. I am connected with an optical audio cable with a decent pair of speakers, and I can tell the difference between 320kbps mp3s and lossless files 100% of the time. I am not exaggerating. I have tested this numerous times and every single time the difference is very apparent for me. Compare that to my $5 headphones. I wouldn't be able to tell lossy vs lossless EVER I'd be willing to bet. So yeah, transparency? Not so much. The only thing "transparent" for me are the lossless files themselves.

I am just wondering if there is at all any difference between a lame encoded 320 which peaks at 20khz and a 320 that peaks at 22khz, and what other lossy encoders are out there besides lame.
Go to the top of the page
+Quote Post
saratoga
post Sep 10 2012, 04:46
Post #4





Group: Members
Posts: 5002
Joined: 2-September 02
Member No.: 3264



You cannot hear such high frequency so no it doesn't matter.

Also transparency isn't subjective and better equipment can make artifacts easier or harder to hear, not just easier like you're assuming.

This post has been edited by saratoga: Sep 10 2012, 04:47
Go to the top of the page
+Quote Post
tpijag
post Sep 10 2012, 05:06
Post #5





Group: Members
Posts: 2352
Joined: 19-May 08
Member No.: 53637



Any reason to wait someone else to google lossy encoders for you?
Go to the top of the page
+Quote Post
xdesirex22
post Sep 10 2012, 06:00
Post #6





Group: Members
Posts: 25
Joined: 20-April 12
Member No.: 99005



QUOTE (saratoga @ Sep 9 2012, 20:46) *
You cannot hear such high frequency so no it doesn't matter.

I'm gonna go ahead and assume you're wrong, just because its you who said it.

QUOTE (saratoga @ Sep 9 2012, 20:46) *
Also transparency isn't subjective

"Transparency, like sound quality, is subjective." source: http://wiki.hydrogenaudio.org/index.php?title=Transparency

QUOTE (saratoga @ Sep 9 2012, 20:46) *
and better equipment can make artifacts easier or harder to hear, not just easier like you're assuming.

I never made this assumption. Feel free to reread my previous post until you fully understand it.

Go to the top of the page
+Quote Post
greynol
post Sep 10 2012, 06:03
Post #7





Group: Super Moderator
Posts: 10009
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (xdesirex22 @ Sep 9 2012, 20:10) *
I can tell the difference between 320kbps mp3s and lossless files 100% of the time.

I don't believe you.

Would you be willing to offer any proof about this in accodrance with our rules?

This post has been edited by greynol: Sep 10 2012, 06:04


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
greynol
post Sep 10 2012, 06:22
Post #8





Group: Super Moderator
Posts: 10009
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (xdesirex22 @ Sep 9 2012, 22:00) *
QUOTE (saratoga @ Sep 9 2012, 20:46) *
You cannot hear such high frequency so no it doesn't matter.

I'm gonna go ahead and assume you're wrong, just because its you who said it.

This behavior is unacceptable.


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
probedb
post Sep 10 2012, 07:54
Post #9





Group: Members
Posts: 1245
Joined: 6-September 04
Member No.: 16817



QUOTE (xdesirex22 @ Sep 10 2012, 04:10) *
I can tell the difference between 320kbps mp3s and lossless files 100% of the time.


Prove it. Where are your ABX results showing this as per the T&Cs you agreed to when you signed up?
Go to the top of the page
+Quote Post
dhromed
post Sep 10 2012, 08:49
Post #10





Group: Members
Posts: 1316
Joined: 16-February 08
From: NL
Member No.: 51347



QUOTE (xdesirex22 @ Sep 10 2012, 05:10) *
I can tell the difference between 320kbps mp3s and lossless files 100% of the time. I am not exaggerating.


No ABX, no dice.
Go to the top of the page
+Quote Post
db1989
post Sep 10 2012, 10:46
Post #11





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



QUOTE (xdesirex22 @ Sep 10 2012, 04:10) *
Not really a fan of the term transparency. First off, it is completely subjective. Different files sound differently to different people.

That’s the point.

Transparency is what is perceptually equivalent to the source for you (any particular “you”), for listening on your equipment.

Also, please don’t come back until you’re willing to comply with rules #8 and #2.
Go to the top of the page
+Quote Post
Kohlrabi
post Sep 10 2012, 11:23
Post #12





Group: Super Moderator
Posts: 1060
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (xdesirex22 @ Sep 10 2012, 05:10) *
I am just wondering if there is at all any difference between a lame encoded 320 which peaks at 20khz and a 320 that peaks at 22khz, and what other lossy encoders are out there besides lame.
First off, I assume that "peaks at" means, "has a lowpass at". I'd guess the file with the higher lowpass at 22kHz might sound worse, because you use the encoder at fixed bitrate, but force it to encode even more mostly inaudible parts of the spectrum, even if the psy-model violently shakes its head against that. Is that assumption valid?

This post has been edited by Kohlrabi: Sep 10 2012, 11:27


--------------------
Ceterum censeo Masterdiskem esse delendam.
Go to the top of the page
+Quote Post
Dynamic
post Sep 10 2012, 18:27
Post #13





Group: Members
Posts: 822
Joined: 17-September 06
Member No.: 35307



Yes, that assumption is probably valid if a 20 kHz lowpass filter is transparent to you*, so long as you stress the word "might" in "might sound worse". In all likelihood they'll sound the same (both be transparent) most of the time given the high bitrate. For problem samples, a few more bits may well be placed "where it matters" and help reduce artifacts if you have a low-pass filter in place.

* pio's public tests at 16kHz lpf give weight to that assumption in the context of real music, not test tones, for a range of listener ages too.

In halb27's lame3.99.5y test encoder, for example, he lowers the lowpass of -V0+ compared to -V0 so he can really "throw the kitchen sink" at encoding short blocks with maximum possible bits without wasting bits on inaudibly high frequencies. It seems to do the trick in fixing short-block problem samples to those people who've tested it so far, and that's well below 20kHz (nearer 17.3 kHz, IIRC), albeit only a couple of semitones lower, and I've not seen that lowpass filter change ABXed versus the original lossless file in any thread I've read. In this case he's doing even better at putting the extra bits "where it matters" by realising that "where it matters" can be further limited to "short blocks" as well as "frequencies up to about 17 kHz".

This post has been edited by Dynamic: Sep 10 2012, 18:30
Go to the top of the page
+Quote Post
yourlord
post Sep 10 2012, 18:49
Post #14





Group: Members
Posts: 219
Joined: 1-March 11
Member No.: 88621



QUOTE (Kohlrabi @ Sep 10 2012, 06:23) *
First off, I assume that "peaks at" means, "has a lowpass at". I'd guess the file with the higher lowpass at 22kHz might sound worse, because you use the encoder at fixed bitrate, but force it to encode even more mostly inaudible parts of the spectrum, even if the psy-model violently shakes its head against that. Is that assumption valid?


You are correct, in a way. Forcing the encoder to consider inaudible content will mean it has to use the fixed bitrate to represent more data than otherwise would be necessary to achieve transparency. That means some of the bit budget is wasted encoding inaudible content when it could have been used to more accurately represent audible content. Having said that, 320Kbps should be so far above the needed bitrate to achieve transparency that on the overwhelming majority of music you'd not hear any difference between the 2. So the quality is essentially the same.

I'll join the chorus wanting ABX results because I simply don't believe anyone who claims they can 100% ABX 320CBR mp3 and lossless on a variety of music, unless they are using a severely broken mp3 encoder.




Go to the top of the page
+Quote Post
greynol
post Sep 10 2012, 18:55
Post #15





Group: Super Moderator
Posts: 10009
Joined: 1-April 04
From: San Francisco
Member No.: 13167



I think we can safely chalk this discussion up as another futile attempt to rank lossy encodings using inappropriate metrics. We don't listen with our eyes.

This post has been edited by greynol: Sep 10 2012, 18:56


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
Dynamic
post Sep 10 2012, 19:15
Post #16





Group: Members
Posts: 822
Joined: 17-September 06
Member No.: 35307



QUOTE (yourlord @ Sep 10 2012, 18:49) *
I'll join the chorus wanting ABX results because I simply don't believe anyone who claims they can 100% ABX 320CBR mp3 and lossless on a variety of music, unless they are using a severely broken mp3 encoder.


Ah, maybe the 22kHz encoder was BLADE? That's certainly a severely broken mp3 encoder that does a fair job at preserving the spectrogram of the original audio, while doing a poor job at preserving the sound. Just right for people who listen with their eyes.
Go to the top of the page
+Quote Post
Gecko
post Sep 10 2012, 19:22
Post #17





Group: Members
Posts: 945
Joined: 15-December 01
From: Germany
Member No.: 662



Also: most of the PA equipment I ever came into contact with didn't waste valuable energy on ultrasonics.

Compared to the HiFi-crowd, the PA-crowd is refreshingly practical and there's far less voodoo.
Go to the top of the page
+Quote Post
slks
post Sep 10 2012, 22:33
Post #18





Group: Members
Posts: 408
Joined: 31-March 06
From: Houston, Texas
Member No.: 29046



To the O.P. - you seem to be one of those who are chasing ever-increasing bit rate and lowpass frequency (or what I liked to call listening by numbers) which is a common mistake. It's a big task to chip away at all the misconceptions that underlie this sort of thinking, but I'm confident that anyone can learn how audio compression works, provided they are willing to learn and think about it scientifically. HA is a great resource for that, since we have rules that any claims of audible difference be backed up by a scientifically valid test (such as ABX). Other forums do not have these rules, and so all sort of misinformation gets passed around unchecked. Besides that, there are quite a few software and codec developers who hang around here, who know their software and the field in general, very deeply. Trying an ABX test of a 320 kb/s LAME encode vs. lossless might be a place to start. You will be surprised to find that unless you've tied the encoder's hands with a list of unnecessary restrictions (no lowpass, forced simple stereo, etc.) you won't be able to tell a difference.

As to your original question-

There is a good reason LAME doesn't encode much of the very high end of the spectrum, and it's because that part of the spectrum is basically inaudible. Lame uses a psymodel (short for psychoacoustic model) to determine what is audible and what isn't, and the most efficient way to allocate the available bits. It takes many things into consideration, but two of these that are particularly relevant-

1. Human hearing sensitivity peaks at 1 or 2 kHz, and drops from there. Many people can't hear anything above ~18 kHz at all, even test tones. By the time you get up to 22 kHz, you're getting into supersonic territory where only bats can hear.

2. There is an effect known as "masking" where quiet sounds at a certain frequency are masked (rendered inaudible) by louder sounds at a different frequency. Similarly to how you won't hear a mosquito buzzing if you're also firing a cannon, although it need not be that extreme of a difference. So even if you can hear a test tone at x kHz, you may not be able to hear sounds at that frequency in the context of actual music, where you've got all kinds of stuff sounding off in other parts of the spectrum.

Now, if you force LAME to encode these inaudible frequencies, it will have less bits to allocate to lower frequency bands that are audible, which could decrease perceptual quality. So what you could potentially end up with is an MP3 that produces a pretty spectrogram, but sounds worse. For MP3s in particular, frequencies above 16 kHz are encoded inefficiently (look up the "sfb21" issue) so forcing the encoder to waste bits in that range is particularly damaging, moreso than it would be for AAC or Vorbis.

320 kb/s is a high enough bit rate that you may not even notice a difference if you force the encoder to go up to 22 kHz. But there's no point in doing so - you gain nothing and potentially lose.


--------------------
http://www.last.fm/user/sls/
Go to the top of the page
+Quote Post
saratoga
post Sep 11 2012, 00:54
Post #19





Group: Members
Posts: 5002
Joined: 2-September 02
Member No.: 3264



QUOTE (xdesirex22 @ Sep 10 2012, 01:00) *
QUOTE (saratoga @ Sep 9 2012, 20:46) *
You cannot hear such high frequency so no it doesn't matter.

I'm gonna go ahead and assume you're wrong, just because its you who said it.


What is that supposed to mean?

Anyway, if you don't believe me you can just check wikipedia and see for yourself.

QUOTE (xdesirex22 @ Sep 10 2012, 01:00) *
QUOTE (saratoga @ Sep 9 2012, 20:46) *
Also transparency isn't subjective

"Transparency, like sound quality, is subjective." source: http://wiki.hydrogenaudio.org/index.php?title=Transparency


Good catch, I fixed that wiki entry so that it doesn't confuse natural variation with subjectivity.

QUOTE (xdesirex22 @ Sep 10 2012, 01:00) *
QUOTE (saratoga @ Sep 9 2012, 20:46) *
and better equipment can make artifacts easier or harder to hear, not just easier like you're assuming.

I never made this assumption. Feel free to reread my previous post until you fully understand it.


"Compare that to my $5 headphones. I wouldn't be able to tell lossy vs lossless EVER I'd be willing to bet."

Many $5 headphones would make spotting artifacts easier, since they would tend to have grossly uneven frequency response. So yes, you are assuming that.
Go to the top of the page
+Quote Post
mjb2006
post Sep 11 2012, 09:07
Post #20





Group: Members
Posts: 816
Joined: 12-May 06
From: Colorado, USA
Member No.: 30694



The post by slks inspired me to do something I've been putting off for a while: write a wiki article that addresses the increasingly common tendency of people to erroneously correlate MP3 quality with high-frequency content and lowpass filter settings.

It's in my usual verbose, over-explanatory, overly technical style. Please feel free to edit it, rename it, add spectrum graphs, etc.: http://wiki.hydrogenaudio.org/index.php?ti...content_in_MP3s
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 2nd October 2014 - 06:22