>>>
This is so misinformed (hence the Slashdot traditional 5)
>>>
I think that has more to do with the fact that it attempts to invalidate a test which shows something Microsoft made in a good daylight and the OS in a bad. Then again, Slashdot does moderate anything remotely sounding correct up, mostly because there is no -1 Wrong moderation.
>>>
Lew Lipnick spent at least a decade as a hi-end audio reviewer
>>>
He didn't set up the test. He only listened. The only way his expertise credits this test is in his listening experience, not in the test setup.
>>>
was apparently accurate enough to clearly differentiate between codecs.
>>>
Where do you get this? There isn't any reliable data about that whatsoever. Please read the journalists response. The test was highly flawed.
>>>
Audio reproduction has come a long way in the last two decades and a modest home system can be very good indeed.
>>>
Uh. An ordinary stereo system is usually budget Japanese consumer electronics. The difference between my Sennheisers and my JVC stereo makes my cry.
>>>
Ummmm, do you think that's why they picked classical and heavy metal sellections? To highlight those differences?
>>>
How does this invalidate the fact that only 2 samples were used?
>>>
If anything, the tacky marketting moniker 'Pro' would bias most audio reviewers against a product. Pro audio gear has long been held in general disdain in hi-end circles as rugged but unmusical.
>>>
Could be true. Then again for software it is usually the reverse, and it were audio codecs they were comparing.
>>>
Rubbing my eyes in disbelief here. Accurate is good! This is what an exciter is for?!?!?
>>>
My mistake. With exciter I'm referring to the kind of 'MP3 enhancer' products that are sold. They just increase the treble a little and boost the basses. Sometimes this makes the music sound better. It doesn't make it more accurate. MP3PRO does just this.
>>>
Double bonk. The sensation of space is created as much by the proper reproduction of room reverberation as it is by directionality. If codecs suppress ambience, a reasonable assumption since they work by discarding low level sound, then yes a space can sound smaller.
>>>
I still remain unconvinced. I agree the spaciousness is dependant on mutiple factors, but I still find it highly suspect they 'thought' Vorbis was worst at this, when this is directly against what would be expected.
I have never experienced this effect in any listening test, although it is hard to compare ithout knowing the bitrate they used.
>>>
How do you know what they've collectively read? Is it impossible they heard this artifact?
>>>
It is not impossible. But it is much more likely
they thought they heard this artefact, because they thoyught they knew how the codec works. It was NOT a blind test.
>>>
Regarding the top end ringing, there's no way to say anything meaningful without knowing what the bit rates were.
>>>
I have to agree with you here.
>>>
I really don't give a rat's ass which codec is superior, I'm just dead tired of seeing misinformation moderated so highly.
>>>
I do give which codes is superior, since I am archiving quite a bit of audio material.
My choice for now is LAME (r3mix settings but
with lower lowpass filer, I can't hear at 19Khz anyway), but when Vorbis 1.0 comes out I will do new listening tests.
>>>
The way I remember it, some of them could distinguish it with significant certainty, though not 100%.
>>>
Do you have a reference for that? I just reread the article and see no such thing, but then again it is in German so it's easier to miss something.
The maximum score was 51 points (perfect recognition), but the maximum that was actually scored was 26 points. This includes the 128Kbps
score. 14,1 points were needed to get a non-guessing score, again including the 128Kbps samples. Considering there were 17 samples I don't see how this can imply that any were able to spot the 256Kbps with any kind of certainty, assuming that if they spot 256Kbps they will also have spot 128Kbps.
Funny is that the best scorer actually had damaged hearing - the psychoacoustics didn't work for him:)
>>>
On the other hand, the comment about loss of spaciousness showed some insight. One of the things most of these lossy formats don't preserve is the phase information between channels.
>>>
I've corrected this elsewhere as well:
The format where they noticed the loss of
spaciousness was the ONLY one that actually
preserved all stereo information. (because it
is not implemented yet)
This is one of the clearest indications their
testing methodology was completely wrong.
Slashdot is not the same uninformed crowd the Washington Post readers are:)
The review was taken down, hung from a tree and shot.
And not because it favored a M$ thingy, but rather
because it was WRONG.
>>>
In any case, after talking to the Post reporter he feels a little sheepish about the whole thing... he thought he balanced the article by mixing positive traits of the openness of the code with a critical quality review and has agreed to be more fair to the first 1.0 encoder release candidate.
>>>
Yes, but did he get the POINT? Or is their next test going to be flawed in exactly the same way?
Also, I hardly noticed any mention of the fact
that mp3pro and wma and proprietary formats that
are unavailable on a lot of systems, whereas
ogg is entirely free...
Monty, when is the rc1 encoder coming out? I hope
REAL soon now?:)
>>>>>
There is a noticable loss of sound quality with any compression technique and IMHO there is no comparison to the original.
>>>>>
Yes, thats what the audiophiles in the c't test (which was correctly conducted) said.
Unfortunately for them, they couldn't distinguish
mp3 @ 256kbps and the cd's AT ALL.
Well, besides the obvious missing details, there
is a lot that is wrong about this:
>>>>>>
The test sessions were done in a home environment with an ordinary stereo system. We focused most of our attention on MP3Pro and Vorbis, the two newest formats, with Windows Audio Media and MP3, the older and more familiar formats, given more limited tests.
>>>>>>
Notice 'a home environment with an ordinary stereo system'. So esentially any more subtle loss in sound quality should have been lost. Great environment for listening tests eh. Note that this isn't compensated by the fact that is what most people use. The distortion between such systems varies widely, and hence what sounds good on one system doesn't necessarily sound good on another.
>>>>>>>
We had them listen to digitally encoded versions of two songs: the opening of a recent recording of Stravinsky's "The Rite of Spring" and the Who's "Love Ain't for Keeping."
>>>>>>>
TWO songs? We have hundreds of music genres and they used two songs for a comparisation? Christ. Encoding Heavy Metal (very bitrate heavy) is a whole different job than encoding classical music(very sensitive for minor distortions).
>>>>>>>
Of the seven listeners, two couldn't discern much difference between MP3Pro and Vorbis. The other five felt Vorbis was the least realistic,
>>>>>>>
Discern difference between the codecs? The way
this paragraph is put makes it highly unlikely
they were doing blind A/B tests.
More likely they actually told the test subjects
which codes it was each time.
This is _always_ going to favor mp3pro., just
because of the name. Also, the point of a good
encoder is to replicate the original music,
not to make it sound good! That is what an
exciter is for.
>>>>
Most thought the beta version of the Vorbis encoder poorly represented the natural sounds of the individual voices or musical instruments.
(A few disagreed, saying certain instruments sounded more synthetic in MP3Pro.)
>>>>
This convices me even more the setup of the test was failed. As I stated in a previous post, mp3pro 'makes up' the high end of the music. This is why some people must have thought it sounded better, while the better trained ones where able to pick up the fact that the high end was artificial.
>>>>
giving higher tones a better ring.
>>>>
Yep. More ringing on the hing end. Sounds like mp3pro for sure. Nevermind that the original music doesnt have it.
>>>>>>
"In the compressed format it sounded as if they had all moved their chairs together," said Hubscher. He founded this especially troublesome in Vorbis.
>>>>>>
BONK. Vorbis is the only format that does NOT
use any kind of joint/intensity stereo coding.
(it will in the 1.0 release)
Then how can it ever get a smaller stereo image??
This isn't making any sense at all...
>>>>>>>
Vrbsky and Lipnick blamed this on the way digital compression shaves off the beginnings and ends of notes.
>>>>>>>
Hahaaaa. They heard something about temporal
masking I'm sure. Too bad they don't have a clue.
>A routing table does not contain a list of all
>the intermediate hops between the address of the
>router and the destination. There is a default
>route and possibly some static routes.
>So unless gnutella creates LOTS of static routes
>in each client, I am not sure how this could
>work.
This is exactly what it does:) I keeps the last few thousand search results it forwarded from those addresses in memory, together with a the address where it got the result from. On a push
request it uses that table to look up where to route the push request to. It does not keep the
entire path from the result sender, just the adress of the node that sent us the result. That node has the address of the node it got the result from, etc...
This is why, if you did a search and got 192.x results, you will sometimes get 'push route no longer available', if one of the hosts inbetween gets dropped between the result receiving and your download attempt. This is also why it's often
so difficult to download from them.
Theres a difference between 192.x addresses on internet and on gnutella. On internet they are unroutable (meant for local networks). On gnutella they just mean 'host behind a firewall, don't try to connect'
>The other thing is, with a large number of
>clients (lime wire being the biggest problem)
>will give you results with the sharing IP as a
>10net or 192.168net address.. these are not
>routable on the net, so you can't even get files
>from them.
They are not routable on the net, but they _are_
routable on gnutella via push messages. Those
addresses basically mean that the client is
behind a firewall and cannot accept connections,
but it _can_ send you the file.
If such a client generates a hit on a search and
sends it result back all clients on the path
between that client and the originator of the
search keep routing information for the 10.x or
192.x address.
If the searcher requests the file it generates
a push message that is sent along the path the
hit came from.
The reason why the 10.x or 192.x addresses are so
unreliable is that many old clients handle them
wrong. If one of those is along the path you
will never get the file, but if all clients along
the path are ok, 10.x/192.x addresses work just
as fine as any other.
The reason why you percieve limewire as more
prone to this prolem is that it is less picky
in allowing connections from older clients, and
hence theres more chance that a bad client is
inbetween a limewire client and yours. But there
is nothing wrong with the limewire client itself.
>Judging by OpenSSH, the OpenBSD team is capable
>of reimplementing complex software in a short
>time span. I wish them luck with the new project.
Uh, they didn't reimplement it.
They took the latest decently licensed version
and improved it from there.
Quite a difference between improving something
and totally rewriting it, even if in the end no
original code remains.
--
GCP
Re:Why it's so small and why you want to avoid it
on
MP3Pro Released
·
· Score: 1
>which is why it can be omitted without immediate
>notice to most listeners.
Up to here I agree. But I am sure even an
untrained listener will be able to pick up a
signal that has been through a 10Khz lowpass
filter. It causes a dull sound, without
definition. Most people will immediately recognize
this as a noticeable quality loss.
It's harder, but very possible to pick out a
16Khz lowpass filter like early MP3 encoders
used, by ear. With 10Khz it's trivial.
MP3Pro tries to hide this via some clever tricks
which make it harder to identify...if you haven't
heard the original recording. Judging by some of
the other posts here there really is a noticable
quality loss, as was to be expected.
Saying this technique gives 'near CD quality' is
the same as saying an audio cassette with Dolby C
is giving 'near CD quality' too. I wanted to
emphasize this fact.
>I think that the psycoacoustic modeling used can
>be even more detrimental to the sound quality.
Yes. Even in the original Mp3 encoders this is
still being improved. LAME for example even
has significant improvements in this regard
with the latest betas (3.88). If you use lame,
experiment a bit with the --athtype 1/2/3 option.
This is good news for projects like Ogg...the
sound quality is already very good now. By
further tuning the psychoacoustics it will only
get better. They are not even using joint stereo
yet!
--
GCP
Re:Why it's so small and why you want to avoid it
on
MP3Pro Released
·
· Score: 1
You need more resolution in the temporal domain.
Actually, its quite more complex than that, since
there are usually fewer high-frequency sounds and
the ATH is higher, but the huffman coding is less
efficient, and probably a few other issues I dont
even know of.
I'm sorry that I can't give you more information,
but this isn't my domain.
Interesting reading on the subject:
http://www.helsinki.fi/~ssyreeni/dsound/dsnda03
--
GCP
Re:Why it's so small and why you want to avoid it
on
MP3Pro Released
·
· Score: 1
>But there is virtually nothing over 15 KHz on
>most music and many sound cards/speakers have
>rolled off response in that range, so we aren't
>missing much.
Thats true. Most (cheap) cards have a little
more distortion and less response in that range,
but it's still quite audible. Do a CD rip, make
a copy of the wav and pass it through a 15Khz
lowpass filter. Now listen to both in sucsession.
On most music it's quite easy to hear the
difference, although having your card connected
to your stereo helps a lot. It also depends on
the CD. Your more likely to hear difference
on DDD (digital recoding/mastering ) CD's than
on AAD (analog recording/mastering) CD's.
Now pass it through a 10Khz lowpass filter. No
matter what equipment you are using, it will
suck. That is the real quality MP3Pro brings you.
The range between 10-15Khz is just 'made up'.
(It's harder to store higher frequencies, hence
by cutting 33% of the frequency range they can
cut bandwidth in half)
--
GCP
Re:Why it's so small and why you want to avoid it
on
MP3Pro Released
·
· Score: 1
Why wait?
You can always listen later to what you encode now.
If 1.0 comes out just upgrade. It's backwards-
compatible.
Also, I was under the impression that any decoder
can decode files from any encoder, but I may have
been wrong.
--
GCP
Re:Comparatively speaking...
on
MP3Pro Released
·
· Score: 5
>CDs are sampled at 128Kbit
Err, hate to tell you this, but you're just plain wrong.
44100 samples/second x 16 bits/sample x 2 channels = 1411200 bits per second
CD's are sampled at 1378 Kbps.
MP3/OGG/WMA can get it down to 128Kbps because
of the compression.
>I can't get Mozilla 0.9.1 because it wants
>Glibc2.2 - which I can't install without putting >at risk my system.
Uhh? Mozilla 0.9.1 works fine with glibc 2.1.
>The Glibc Heck has proven to be quite a pain in
>the hard drive. If anyone has a solution that
>doesn't involve rebuilding the whole packages (I
>already thought about that, thx), I'd be glad to
>hear.
There's no good solution that I know of. Upgrading
from libc5 to glibc2 was a major pain. The easiest
way is to install a new distribution.
Glibc 'upgrades' happen very infrequently just
because of this reason (libc5->glibc2 was the
only one I went through and I've been using Linux
since 1.2.13) The main reason for that upgrade
was localization and threads support. If you
didn't need either of those, libc5 is still fine.
AFAIK there is nothing that needs a glibc2.2 right now.
--
GCP
Why it's so small and why you want to avoid it
on
MP3Pro Released
·
· Score: 5
If you like your records, MP3Pro is something to
stay away from.
It attains such a high compression by using a
technique of constructing the higher frequencies
by _guessing_ what the ones that the compression
left out where, based on the lower frequencies,
and amplifying the rest.
You could compare this to saying that a cassette
sounds just as good as a CD if you just use
Dobly B/C. Not.
MP3Pro is limited to 10Khz, and can replicate
the sounds up to 15Khz. A cd is 22Khz and the
human ear can go to 19Khz for a normal healty
person. This means that you LOSE over half
the spectrum. Sure, you may not notice it
immediately because of the 'guessing' and the
'replictation', but if will be gruesome when
compared to the original CD.
Face it, you can't do wonders AND stay compatible
with old mp3 players.
Sure, it's a nice trick for streaming if 64Kbps
is all you have, but it's not fundamentally
different from the old mp3 format and using an
exciter plugin. The utility is severly limited.
That said, just use Ogg. It works. Yes, I really
mean that. The sound quality is great, the tools
are stable enough (beta4), and plugins are available
for most importants apps.
All it's missing is an ACM plugin for Windows so
non-Ogg-aware can deal with it too. Not that there
are many left. All serious sound editing packages
have native support now. And yes, it's being worked
on.
>when most isp's/cable/dsl providers prohibit you
>from running servers?
Good point. There was serious discussion some time
ago on the mailinglist about techniques to allow
setting a node to only accept connections from
trusted nodes. That way it would be impossible
for someone to detect that you were running a
freenet node from the outside, while you still
had full access to the network and your node
could be used for storage.
It was turned down by Ian because he thought it
was a non-issue.
The argument used was that once Freenet becomes
popular enough there is no reason why running
a Freenet node means that you are doing something
suspicious. And Freenet grows easier with fully
operative nodes than with those 'stealth' nodes.
(at least that's what I remember from the discussion)
The problem of course is, that Freenet will never
become popular if ISP's start shutting the servers
down.
Ian's reply to this was:
'It is my experience that when users demand
P2P and Freenet access all ISP's will bend
over backwards to give it to them'
And I think he may be right about that. My own
ISP interpreted the 'server' clause in a way
that you were free to use napster as long as
you set the number of allowed incoming connections
to zero (effectively disabling the server). That
way people cannot upload from you, which is what
was actually illegal about Napster (downloading
is fine as you might own the CD yourself)
The ISP WANTS to offer Napster to people. They
offer broanband services so Napster is a good
reason to switch over from PPP. So they make
sure the customer CAN have Napster.
Slashdot Mirror
>As a matter of fact, mp3 is a compression scheme
>for wav format. So it _must_ be converted from
>wav.
This is simply wrong. mp3 compresses PCM audio
data. It being in RIFF WAVE format has nothing
to do with mp3.
--
GCP
>a Lame VBR mp3 is higher quality than an Ogg
>file anyway.
This is debatable...certainly _not_ for equal
average resulting bitrates.
>But in the consumer market, MP3 was there first,
>MP3 is already popular.. and it's another VHS
>versus Betamax.
That would be good. VHS won because it was more
usable and was a more open format. Vorbis has all
this and better quality.
--
GCP
>1. Lack of portable hardware players. All the
>players on the market today support mp3 and wma,
>but none play ogg. This is a problem.
Iomega HipZip does, others are comming...
--
GCP
>>>
This is so misinformed (hence the Slashdot traditional 5)
>>>
I think that has more to do with the fact that it attempts to invalidate a test which shows something Microsoft made in a good daylight and the OS in a bad. Then again, Slashdot does moderate anything remotely sounding correct up, mostly because there is no -1 Wrong moderation.
>>>
Lew Lipnick spent at least a decade as a hi-end audio reviewer
>>>
He didn't set up the test. He only listened. The only way his expertise credits this test is in his listening experience, not in the test setup.
>>>
was apparently accurate enough to clearly differentiate between codecs.
>>>
Where do you get this? There isn't any reliable data about that whatsoever. Please read the journalists response. The test was highly flawed.
>>>
Audio reproduction has come a long way in the last two decades and a modest home system can be very good indeed.
>>>
Uh. An ordinary stereo system is usually budget Japanese consumer electronics. The difference between my Sennheisers and my JVC stereo makes my cry.
>>>
Ummmm, do you think that's why they picked classical and heavy metal sellections? To highlight those differences?
>>>
How does this invalidate the fact that only 2 samples were used?
>>>
If anything, the tacky marketting moniker 'Pro' would bias most audio reviewers against a product. Pro audio gear has long been held in general disdain in hi-end circles as rugged but unmusical.
>>>
Could be true. Then again for software it is usually the reverse, and it were audio codecs they were comparing.
>>>
Rubbing my eyes in disbelief here. Accurate is good! This is what an exciter is for?!?!?
>>>
My mistake. With exciter I'm referring to the kind of 'MP3 enhancer' products that are sold. They just increase the treble a little and boost the basses. Sometimes this makes the music sound better. It doesn't make it more accurate. MP3PRO does just this.
>>>
Double bonk. The sensation of space is created as much by the proper reproduction of room reverberation as it is by directionality. If codecs suppress ambience, a reasonable assumption since they work by discarding low level sound, then yes a space can sound smaller.
>>>
I still remain unconvinced. I agree the spaciousness is dependant on mutiple factors, but I still find it highly suspect they 'thought' Vorbis was worst at this, when this is directly against what would be expected.
I have never experienced this effect in any listening test, although it is hard to compare ithout knowing the bitrate they used.
>>>
How do you know what they've collectively read? Is it impossible they heard this artifact?
>>>
It is not impossible. But it is much more likely
they thought they heard this artefact, because they thoyught they knew how the codec works. It was NOT a blind test.
>>>
Regarding the top end ringing, there's no way to say anything meaningful without knowing what the bit rates were.
>>>
I have to agree with you here.
>>>
I really don't give a rat's ass which codec is superior, I'm just dead tired of seeing misinformation moderated so highly.
>>>
I do give which codes is superior, since I am archiving quite a bit of audio material.
My choice for now is LAME (r3mix settings but
with lower lowpass filer, I can't hear at 19Khz anyway), but when Vorbis 1.0 comes out I will do new listening tests.
--
GCP
>>>
:)
The way I remember it, some of them could distinguish it with significant certainty, though not 100%.
>>>
Do you have a reference for that? I just reread the article and see no such thing, but then again it is in German so it's easier to miss something.
The maximum score was 51 points (perfect recognition), but the maximum that was actually scored was 26 points. This includes the 128Kbps
score. 14,1 points were needed to get a non-guessing score, again including the 128Kbps samples. Considering there were 17 samples I don't see how this can imply that any were able to spot the 256Kbps with any kind of certainty, assuming that if they spot 256Kbps they will also have spot 128Kbps.
Funny is that the best scorer actually had damaged hearing - the psychoacoustics didn't work for him
--
GCP
>>>
On the other hand, the comment about loss of spaciousness showed some insight. One of the things most of these lossy formats don't preserve is the phase information between channels.
>>>
I've corrected this elsewhere as well:
The format where they noticed the loss of
spaciousness was the ONLY one that actually
preserved all stereo information. (because it
is not implemented yet)
This is one of the clearest indications their
testing methodology was completely wrong.
--
GCP
>>>
:)
:)
:)
and Slashdot covers that one.
>>>
Slashdot is not the same uninformed crowd the Washington Post readers are
The review was taken down, hung from a tree and shot.
And not because it favored a M$ thingy, but rather
because it was WRONG.
>>>
In any case, after talking to the Post reporter he feels a little sheepish about the whole thing... he thought he balanced the article by mixing positive traits of the openness of the code with a critical quality review and has agreed to be more fair to the first 1.0 encoder release candidate.
>>>
Yes, but did he get the POINT? Or is their next test going to be flawed in exactly the same way?
Also, I hardly noticed any mention of the fact
that mp3pro and wma and proprietary formats that
are unavailable on a lot of systems, whereas
ogg is entirely free...
Monty, when is the rc1 encoder coming out? I hope
REAL soon now?
--
GCP (+1 to parent, monty rocks!
If the tests are properly conducted, this is simply not true.
--
GCP
>>>
Play on my stereo an MP3 vs. a CD of the same song I will 100% pick out the CD as being better each time
>>>
At what bitrate?
I'd be really really really surprised if you
could do this for 256Kps MP3 with a recent
encoder (LAME 3.88 or Frauenhofer even)
As for 128Kps, no doubt. I can do it too,
but not for all songs (and with headphones
only)
--
GCP
>>> :)
He probably heard some flanging artifact and thought it was a valve closure
>>>
Exactly my thought
This is why you really must do blind A/B tests.
>>>
(Credit to them grokking that lossy compression in music often throws away stereo separation / spatial components, though.)
>>>
Grokking? Nah, more like: 'lets try to sound smart'
They managed to blame the sole codec that does NOT use intensity/joint stereo of losing stereo definition.
Which nicely sums up the reliabily of their test...
--
GCP
>>>>>
There is a noticable loss of sound quality with any compression technique and IMHO there is no comparison to the original.
>>>>>
Yes, thats what the audiophiles in the c't test (which was correctly conducted) said.
Unfortunately for them, they couldn't distinguish
mp3 @ 256kbps and the cd's AT ALL.
--
GCP
Well, besides the obvious missing details, there
is a lot that is wrong about this:
>>>>>>
The test sessions were done in a home environment with an ordinary stereo system. We focused most of our attention on MP3Pro and Vorbis, the two newest formats, with Windows Audio Media and MP3, the older and more familiar formats, given more limited tests.
>>>>>>
Notice 'a home environment with an ordinary stereo system'. So esentially any more subtle loss in sound quality should have been lost. Great environment for listening tests eh. Note that this isn't compensated by the fact that is what most people use. The distortion between such systems varies widely, and hence what sounds good on one system doesn't necessarily sound good on another.
>>>>>>>
We had them listen to digitally encoded versions of two songs: the opening of a recent recording of Stravinsky's "The Rite of Spring" and the Who's "Love Ain't for Keeping."
>>>>>>>
TWO songs? We have hundreds of music genres and they used two songs for a comparisation? Christ. Encoding Heavy Metal (very bitrate heavy) is a whole different job than encoding classical music(very sensitive for minor distortions).
>>>>>>>
Of the seven listeners, two couldn't discern much difference between MP3Pro and Vorbis. The other five felt Vorbis was the least realistic,
>>>>>>>
Discern difference between the codecs? The way
this paragraph is put makes it highly unlikely
they were doing blind A/B tests.
More likely they actually told the test subjects
which codes it was each time.
This is _always_ going to favor mp3pro., just
because of the name. Also, the point of a good
encoder is to replicate the original music,
not to make it sound good! That is what an
exciter is for.
>>>>
Most thought the beta version of the Vorbis encoder poorly represented the natural sounds of the individual voices or musical instruments.
(A few disagreed, saying certain instruments sounded more synthetic in MP3Pro.)
>>>>
This convices me even more the setup of the test was failed. As I stated in a previous post, mp3pro 'makes up' the high end of the music. This is why some people must have thought it sounded better, while the better trained ones where able to pick up the fact that the high end was artificial.
>>>>
giving higher tones a better ring.
>>>>
Yep. More ringing on the hing end. Sounds like mp3pro for sure. Nevermind that the original music doesnt have it.
>>>>>>
"In the compressed format it sounded as if they had all moved their chairs together," said Hubscher. He founded this especially troublesome in Vorbis.
>>>>>>
BONK. Vorbis is the only format that does NOT
use any kind of joint/intensity stereo coding.
(it will in the 1.0 release)
Then how can it ever get a smaller stereo image??
This isn't making any sense at all...
>>>>>>>
Vrbsky and Lipnick blamed this on the way digital compression shaves off the beginnings and ends of notes.
>>>>>>>
Hahaaaa. They heard something about temporal
masking I'm sure. Too bad they don't have a clue.
--
GCP (who did his own listening tests)
>A routing table does not contain a list of all
:) I keeps the last few thousand search results it forwarded from those addresses in memory, together with a the address where it got the result from. On a push
>the intermediate hops between the address of the
>router and the destination. There is a default
>route and possibly some static routes.
>So unless gnutella creates LOTS of static routes
>in each client, I am not sure how this could
>work.
This is exactly what it does
request it uses that table to look up where to route the push request to. It does not keep the
entire path from the result sender, just the adress of the node that sent us the result. That node has the address of the node it got the result from, etc...
This is why, if you did a search and got 192.x results, you will sometimes get 'push route no longer available', if one of the hosts inbetween gets dropped between the result receiving and your download attempt. This is also why it's often
so difficult to download from them.
Theres a difference between 192.x addresses on internet and on gnutella. On internet they are unroutable (meant for local networks). On gnutella they just mean 'host behind a firewall, don't try to connect'
--
GCP
>The other thing is, with a large number of
>clients (lime wire being the biggest problem)
>will give you results with the sharing IP as a
>10net or 192.168net address.. these are not
>routable on the net, so you can't even get files
>from them.
They are not routable on the net, but they _are_
routable on gnutella via push messages. Those
addresses basically mean that the client is
behind a firewall and cannot accept connections,
but it _can_ send you the file.
If such a client generates a hit on a search and
sends it result back all clients on the path
between that client and the originator of the
search keep routing information for the 10.x or
192.x address.
If the searcher requests the file it generates
a push message that is sent along the path the
hit came from.
The reason why the 10.x or 192.x addresses are so
unreliable is that many old clients handle them
wrong. If one of those is along the path you
will never get the file, but if all clients along
the path are ok, 10.x/192.x addresses work just
as fine as any other.
The reason why you percieve limewire as more
prone to this prolem is that it is less picky
in allowing connections from older clients, and
hence theres more chance that a bad client is
inbetween a limewire client and yours. But there
is nothing wrong with the limewire client itself.
--
GCP
Please reread my post.
I never mentioned ipf/pf. I was talking about OpenSSH.
--
GCP
>Judging by OpenSSH, the OpenBSD team is capable
>of reimplementing complex software in a short
>time span. I wish them luck with the new project.
Uh, they didn't reimplement it.
They took the latest decently licensed version
and improved it from there.
Quite a difference between improving something
and totally rewriting it, even if in the end no
original code remains.
--
GCP
>which is why it can be omitted without immediate
>notice to most listeners.
Up to here I agree. But I am sure even an
untrained listener will be able to pick up a
signal that has been through a 10Khz lowpass
filter. It causes a dull sound, without
definition. Most people will immediately recognize
this as a noticeable quality loss.
It's harder, but very possible to pick out a
16Khz lowpass filter like early MP3 encoders
used, by ear. With 10Khz it's trivial.
MP3Pro tries to hide this via some clever tricks
which make it harder to identify...if you haven't
heard the original recording. Judging by some of
the other posts here there really is a noticable
quality loss, as was to be expected.
Saying this technique gives 'near CD quality' is
the same as saying an audio cassette with Dolby C
is giving 'near CD quality' too. I wanted to
emphasize this fact.
>I think that the psycoacoustic modeling used can
>be even more detrimental to the sound quality.
Yes. Even in the original Mp3 encoders this is
still being improved. LAME for example even
has significant improvements in this regard
with the latest betas (3.88). If you use lame,
experiment a bit with the --athtype 1/2/3 option.
This is good news for projects like Ogg...the
sound quality is already very good now. By
further tuning the psychoacoustics it will only
get better. They are not even using joint stereo
yet!
--
GCP
You need more resolution in the temporal domain.
Actually, its quite more complex than that, since
there are usually fewer high-frequency sounds and
the ATH is higher, but the huffman coding is less
efficient, and probably a few other issues I dont
even know of.
I'm sorry that I can't give you more information,
but this isn't my domain.
Interesting reading on the subject:
http://www.helsinki.fi/~ssyreeni/dsound/dsnda03
--
GCP
>But there is virtually nothing over 15 KHz on
>most music and many sound cards/speakers have
>rolled off response in that range, so we aren't
>missing much.
Thats true. Most (cheap) cards have a little
more distortion and less response in that range,
but it's still quite audible. Do a CD rip, make
a copy of the wav and pass it through a 15Khz
lowpass filter. Now listen to both in sucsession.
On most music it's quite easy to hear the
difference, although having your card connected
to your stereo helps a lot. It also depends on
the CD. Your more likely to hear difference
on DDD (digital recoding/mastering ) CD's than
on AAD (analog recording/mastering) CD's.
Now pass it through a 10Khz lowpass filter. No
matter what equipment you are using, it will
suck. That is the real quality MP3Pro brings you.
The range between 10-15Khz is just 'made up'.
(It's harder to store higher frequencies, hence
by cutting 33% of the frequency range they can
cut bandwidth in half)
--
GCP
Why wait?
You can always listen later to what you encode now.
If 1.0 comes out just upgrade. It's backwards-
compatible.
Also, I was under the impression that any decoder
can decode files from any encoder, but I may have
been wrong.
--
GCP
>CDs are sampled at 128Kbit
Err, hate to tell you this, but you're just plain wrong.
44100 samples/second x 16 bits/sample x 2 channels = 1411200 bits per second
CD's are sampled at 1378 Kbps.
MP3/OGG/WMA can get it down to 128Kbps because
of the compression.
--
GCP
>I can't get Mozilla 0.9.1 because it wants
>Glibc2.2 - which I can't install without putting >at risk my system.
Uhh? Mozilla 0.9.1 works fine with glibc 2.1.
>The Glibc Heck has proven to be quite a pain in
>the hard drive. If anyone has a solution that
>doesn't involve rebuilding the whole packages (I
>already thought about that, thx), I'd be glad to
>hear.
There's no good solution that I know of. Upgrading
from libc5 to glibc2 was a major pain. The easiest
way is to install a new distribution.
Glibc 'upgrades' happen very infrequently just
because of this reason (libc5->glibc2 was the
only one I went through and I've been using Linux
since 1.2.13) The main reason for that upgrade
was localization and threads support. If you
didn't need either of those, libc5 is still fine.
AFAIK there is nothing that needs a glibc2.2 right now.
--
GCP
If you like your records, MP3Pro is something to
stay away from.
It attains such a high compression by using a
technique of constructing the higher frequencies
by _guessing_ what the ones that the compression
left out where, based on the lower frequencies,
and amplifying the rest.
You could compare this to saying that a cassette
sounds just as good as a CD if you just use
Dobly B/C. Not.
MP3Pro is limited to 10Khz, and can replicate
the sounds up to 15Khz. A cd is 22Khz and the
human ear can go to 19Khz for a normal healty
person. This means that you LOSE over half
the spectrum. Sure, you may not notice it
immediately because of the 'guessing' and the
'replictation', but if will be gruesome when
compared to the original CD.
Face it, you can't do wonders AND stay compatible
with old mp3 players.
Sure, it's a nice trick for streaming if 64Kbps
is all you have, but it's not fundamentally
different from the old mp3 format and using an
exciter plugin. The utility is severly limited.
That said, just use Ogg. It works. Yes, I really
mean that. The sound quality is great, the tools
are stable enough (beta4), and plugins are available
for most importants apps.
All it's missing is an ACM plugin for Windows so
non-Ogg-aware can deal with it too. Not that there
are many left. All serious sound editing packages
have native support now. And yes, it's being worked
on.
--
GCP
>when most isp's/cable/dsl providers prohibit you
>from running servers?
Good point. There was serious discussion some time
ago on the mailinglist about techniques to allow
setting a node to only accept connections from
trusted nodes. That way it would be impossible
for someone to detect that you were running a
freenet node from the outside, while you still
had full access to the network and your node
could be used for storage.
It was turned down by Ian because he thought it
was a non-issue.
The argument used was that once Freenet becomes
popular enough there is no reason why running
a Freenet node means that you are doing something
suspicious. And Freenet grows easier with fully
operative nodes than with those 'stealth' nodes.
(at least that's what I remember from the discussion)
The problem of course is, that Freenet will never
become popular if ISP's start shutting the servers
down.
Ian's reply to this was:
'It is my experience that when users demand
P2P and Freenet access all ISP's will bend
over backwards to give it to them'
And I think he may be right about that. My own
ISP interpreted the 'server' clause in a way
that you were free to use napster as long as
you set the number of allowed incoming connections
to zero (effectively disabling the server). That
way people cannot upload from you, which is what
was actually illegal about Napster (downloading
is fine as you might own the CD yourself)
The ISP WANTS to offer Napster to people. They
offer broanband services so Napster is a good
reason to switch over from PPP. So they make
sure the customer CAN have Napster.
--
GCP
How will this affect the future of Mozilla?
Yes, its under a free license, but let's not
forget nearly all development is still done
by Netscape employees.
If 80% of the developers have to work on other
stuff, it's going to be Nomorezilla fast enough...
--
GCP