IPB

Welcome Guest ( Log In | Register )

8 Pages V  « < 2 3 4 5 6 > »   
Reply to this topicStart new topic
HELP! Please URGENT (mp3 vs cd-audio comparison)
Pio2001
post Dec 4 2002, 13:23
Post #76


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



Discussion about CD life expectancy is splitted here : http://www.hydrogenaudio.org/forums/index....=ST&f=20&t=4673
Go to the top of the page
+Quote Post
JohnV
post Dec 4 2002, 13:52
Post #77





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



budgie:
Here are 5 samples. Please tell me which one is the original, and which one is the "crappy mp3" and which one is the second best... So there's only 1 original and 4 lossy audio samples.
http://static.hydrogenaudio.org/extra/test/

Sample is quite hard, so this should be relatively easy test. wink.gif And use only your ears..

This is only for budgie, if somebody else wants to comment about quality of these samples, please do it via PM. I know this is not quite valid test, since there's no reference, but if the mp3 is so crappy, surely it must be recognaized.


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
2Bdecided
post Dec 4 2002, 14:16
Post #78


ReplayGain developer


Group: Developer
Posts: 5066
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



budgie seems to think we're just wasting our time, and changing the "sound" of good music for no apparent reason.

1. does MPC change the sound? prove it!

2. the reason is to fit all the music I own on a hard disk drive. With only lossless compression, the whole thing is unmanagable. With lossy MPC, it's possible.

D.
Go to the top of the page
+Quote Post
NumLOCK
post Dec 4 2002, 14:30
Post #79


Neutrino G-RSA developer


Group: Developer
Posts: 852
Joined: 8-May 02
From: Geneva
Member No.: 2002



budgie =>

QUOTE
Oh God, so you want to tell me you know more than 1 man who definitely can hear up to 20 kHz? Do you know something about the matter?! ohmy.gif Why I have more and more stronger feeling I've turned out to be in a very weird place? Everybody round me is a well trained genius with GOLDEN EARS and me, poor dumb, deaf man, who just happened to live through the past 20 years as a musician and sound engineer?

You got me wrong (thanks Kblood and Pio2001) - I never pretended that any human could hear anything up to 20 kHz. Please read my post again !
By the way I didn't talk about myself at all - I was simply mentioning 18.5 and 19.5kHz, as two lowpass filter values that you can use for high-quality mp3 encoding.

Apart from that, I said that unless I'm mistaken, the --alt-preset standard preset applies a 18.5kHz lowpass to the source. Use a.p.extreme for 19.5kHz.

As an anecdote you can also disable the lowpass, but:
- it usually introduces twinkling issues as well as aliasing,
- encoding the (ultrasonic) noise that's between 20-22kHz, is usually just a waste of space.

As you know, the ear's limits can come quite close to 20 kHz. I could even, myself, distinguish between a music sample lowpassed @ 18.5kHz and the same sample lowpassed @ 19.5 kHz. The properties of the filter might have come in play here. As expected, that (very very slight) difference was found on harsh transients with significant harmonic content.

With a sine tone, last time I checked my ears did cutoff near ~19kHz. I know that's not very common, even though I was 19 years old by then - and no, it wasn't on my PC.

You're not in a weird place, budgie. You're just at a crossroad between traditional audiophile people and cutting-edge, Plextor-trusting audio freaks. A good compromise if you prefer.
By the way, you're welcome in this forum, even if we don't agree on everything.

Again, you being a musician doesn't imply that you're trained yourself for the same type of listening as a codec tester !
The opposite is true: codec testers don't need to know music well.
After much discussion I had with musicians (including a "chef d'orchestre"), it was obvious that we didn't approach music listening from the same perspective ! But what was really interesting, was the cooperation between people with different points of view. It's the only way to go further.

We came up with a few interesting points, and after some tests we found they couldn't distinguish a properly-encoded ~170kbps VBR file with the original CD - no matter how hard they tried. They had no special training before the test.

After all, while you're trying to find defects on the original CD, well, some people here take your CD for a perfect source, and they're trying to make audio codecs fail with it ! Same music, yes, but different approaches really !

Perhaps this question might help us: I know you don't perceive mp3 and cd the same way. Whether placebo or not, I don't know yet. But have you already encoded, say, a good quality mp3 and heard an obvious artifact in it ?

If so, you'll soon see that the best way to isolate the problem is to cut the 2 samples... to listen to both... to randomize the order of the listening... and what do you get next ? ABX testing smile.gif

This post has been edited by NumLOCK: Dec 4 2002, 15:43


--------------------
Try Leeloo Chat at http://leeloo.webhop.net
Go to the top of the page
+Quote Post
mithrandir
post Dec 4 2002, 19:26
Post #80





Group: Members
Posts: 669
Joined: 15-January 02
From: SE Pennsylvania
Member No.: 1032



Most humans can hear over 20KHz. It's just that our ears are very insensitive in that frequency region leading us to say we can't hear that high. But if any pure signal in the 20Hz-22KHz band is strong enough, we can hear it. I'm sure most would be able to hear a 20KHz signal recorded at -5dB. However, such HF signals are almost never that loud in recorded music.

With most music, if you simply cut off all information above 16KHz, 17KHz, 18KHz, etc. it seems most listeners can't tell a single difference. That's because so much musical information is anchored in the middle frequency ranges, which subsequently help mask quiet, very high frequency information.

A fixed frequency lowpass is not a very good solution. If you set the lowpass at, say, 18.5KHz, you automatically eliminate 20KHz signals, even if they are very, very strong. A better solution is to use a properly adjusted ATH profile for the entire frequency spectrum. MPC does it this way and I believe Vorbis does as well. LAME employs a lowpass because of MP3's sfb21 missing scalefactor. You don't want too much information stored in this sfb or else you won't be able to use a scalefactor at all, a practice which can lead to significant bit bloat.

This post has been edited by mithrandir: Dec 4 2002, 19:28
Go to the top of the page
+Quote Post
Pio2001
post Dec 4 2002, 20:39
Post #81


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



I can hear up to 16 kHz, but I am unable to hear a 100 db 18 kHz pure sine. Hearing ability decrease very fast with frequency, ater all, only one tone separates 16 kHz from 18 kHz.

This post has been edited by Pio2001: Dec 4 2002, 20:40
Go to the top of the page
+Quote Post
SLOQshtr
post Dec 4 2002, 20:45
Post #82





Group: Members
Posts: 11
Joined: 1-December 02
Member No.: 3939



Good to see this discusion is going on.

I still haven't got that CD. I hang out here every day. As I see most of opinions are in my favor which is good to hear.

After the test I plan to put the short results here, and maybe longer somewhere else on the web.

Thanx everyone for their input.

qshtr
Go to the top of the page
+Quote Post
mmortal03
post Dec 5 2002, 00:34
Post #83





Group: Members
Posts: 601
Joined: 19-July 02
From: USA
Member No.: 2667



One question of budgie that has not been answered yet was questioning the point in having MP3 in the first place, where he under-emphasized the format as only being useful for low bit-rate portability. There is a HUGE range of possibilites in mp3, and EACH has its own possible application, even archiving, because each person has their own perception of what makes something useful for them. Regardless of whether one agrees with them or not because of one's higher standards, some people just don't care, and are happy with what they have; not to downgrade the quality of mp3. File sharing is one major reason that mp3's compression is so useful. Even if one does not feel the encoding is adequate for archiving, it is still often listenable and he or she can then go buy the cd or download a higher quality encoding. Bands are discovered by listeners everyday on their computer, and it is mainly because of mp3. I eventually buy anything that I find that I like, so there is nothing lost by it. Bandwidth and storage ties right in to Filesharing. ANY gain of filesize can be significant to SOMEONE. Even at --alt-preset standard, I can fit many albums on one cd and listen to it on my mp3-cd player and the quality is great. Albums that I have stored on my computer may not be perfect copies persay, but they are much more easily searched and played in this format compared to the physical cd format with the normal person only having one to two cd compatible drives on his or her computer. This makes my normal use of my cds far less (until my mp3cd player broke the other day) and makes them last longer. You might say that my CDs are still my music "archived" and my mp3s are my most used music. The 2nd definition in computer science of the noun "archive" on dictionary.com may not be known by some, and it is: A file containing one or more files in compressed format for more efficient storage and transfer. While this usually pertains to zip files or other file compression formats, Mp3 does that very well as well, and since it is "not the best", there are other formats to choose from. Isn't there a thing called preference?


Also, A question for mithrandir: What is usually lost by cutting the signal off at 20Hz-22KHz? What I mean by "lost" is, what natural instrumentation or sound requires that high of frequency to be perceived as it was meant to be heard? As you said, not many things are recorded 20KHz or higher purposely, because most of the music that is meant to be heard falls in the middle ranges. When a musician is recorded with the proper equipment, I assume that more sound range information is recorded than what is required for his particular music or instrumentation, and even some of that unneeded information is then downsampled onto his CD when it is finished (this is overly simplified simply because I do not intend to get into the inner workings of studio sound recording at the moment, that is not my concern.) Because of this, I want to know what types of music and instrumentation recorded out there makes use of or contains significant sound information in these higher frequencies. People talk about how they can hear in those ranges or not, but if you can't, what are you missing, and is it significant? I know that significance is always opinionated, and I'd like to hear opinions as well on this.

This post has been edited by mmortal03: Dec 5 2002, 00:39


--------------------
WARNING: Changing of advanced parameters might degrade sound quality. Modify them only if you are expirienced in audio compression!
Go to the top of the page
+Quote Post
mp3fan
post Dec 5 2002, 00:47
Post #84





Group: Members
Posts: 94
Joined: 13-October 01
Member No.: 281



It would be highly unlikely that someone here can use ff123's website test and say that they can hear up to 20khz. Someone said earlier that most humans can hear up to 20KHz? That's false. Sensitivity matters. I don't qualify someone as being able to hear 19KHz if it requires 120db of loudness to do so. And that's EXACTLY what "most" humans would need to perceive such a tone.

In short, any tone above 17KHz (in my personal experiments of others) is inaudible and therefore I've concluded with that a lowpass of even 18KHz is wasteful.

I believe that we should be exploring the tuning of the -Y switch to cut down bitrate and find the correct sensitivity using ABX tests. This would help achieve what VBR was designed for, and that's to have the best possible sound at the lowest possible bitrate.

mp3
Go to the top of the page
+Quote Post
mithrandir
post Dec 5 2002, 03:06
Post #85





Group: Members
Posts: 669
Joined: 15-January 02
From: SE Pennsylvania
Member No.: 1032



QUOTE (mmortal03 @ Dec 4 2002 - 06:34 PM)
Also,  A question for mithrandir:  What is usually lost by cutting the signal off at 20Hz-22KHz?  What I mean by "lost" is, what natural instrumentation or sound requires that high of frequency to be perceived as it was meant to be heard?  As you said, not many things are recorded 20KHz or higher purposely, because most of the music that is meant to be heard falls in the middle ranges.  When a musician is recorded with the proper equipment, I assume that more sound range information is recorded than what is required for his particular music or instrumentation, and even some of that unneeded information is then downsampled onto his CD when it is finished (this is overly simplified simply because I do not intend to get into the inner workings of studio sound recording at the moment, that is not my concern.)  Because of this, I want to know what types of music and instrumentation recorded out there makes use of or contains significant sound information in these higher frequencies.  People talk about how they can hear in those ranges or not, but if you can't, what are you missing, and is it significant?  I know that significance is always opinionated, and I'd like to hear opinions as well on this.

In theory you want to encode both normal and ultrasonic frequencies. Some instrumentation might produce energy as high as 100KHz. I guess that's one reason why SACD and DVD-A increase the upper end of the recorded spectrum from 22KHz to 48KHz. Why do you want to encode such high frequencies? Because they exist in live performances, mostly as harmonics of lower frequencies. Whether the inclusion of these ultrasonic frequencies makes an impact on replayed musical events is subject to debate. I'm comfortable with having normal music rolled off well before 20KHz when using a lossy format. The whole point of lossy is getting rid of data you can't perceive anyway. I'm not sure I could tell the difference between 44.1KHz sampled and 96KHz sampled music. But if I was using a lossless format, I'd want to store the entire frequency spectrum.

The usuable dynamic range for 16-bit digital recording is about 95-100dB: the quietest signals on a CD are quite a bit down in level from the loudest. In some live musical events, I'd estimate that information in the 22KHz+ range would be strong enough to be retained by a 16-bit recording. Is it useful information? I don't know. The new 24-bit formats are marketted as if ultrasonic frequencies are important and even if they are, most playback equipment is insufficient. My loudspeakers are designed for flat response from about 60Hz-18KHz. I'm sure the tweeter is well down in level by 30KHz and probably suffers from ringing effects above 25KHz or so like most tweeters made for 44.1KHz digital. So maybe if playback equipment were perfect, you'd want to retain the ultrasonic information but the payoffs in sonic quality seem far larger by concentrating on the midrange. Get rid of boxiness, nasality, chestiness, woodenness, etc. before spending time and money on the ultrasonics.
Go to the top of the page
+Quote Post
mithrandir
post Dec 5 2002, 03:14
Post #86





Group: Members
Posts: 669
Joined: 15-January 02
From: SE Pennsylvania
Member No.: 1032



QUOTE (mp3fan @ Dec 4 2002 - 06:47 PM)
In short, any tone above 17KHz (in my personal experiments of others) is inaudible and therefore I've concluded with that a lowpass of even 18KHz is wasteful.

I believe that we should be exploring the tuning of the -Y switch to cut down bitrate and find the correct sensitivity using ABX tests.  This would help achieve what VBR was designed for, and that's to have the best possible sound at the lowest possible bitrate.

mp3

Well, that's not true. I used to be able to (and many still can) clearly hear the 19KHz signal given off by most television sets. And when I was younger I could very well hear 20KHz tones on those test discs.

But those tones were mostly unmasked. In normal music, most 17KHz+ signals aren't "important", important being defined as audible on an ABX test. There are a few lowpass tests out there and I've surprised myself in one case where I couldn't tell the difference between the original and a 14KHz lowpassed file. But I (and just about everyone else) can hear information above 14KHz. Masking and relative volume have much to do with this.

I like using -Y (or --ns-sfb) with APS to really cut down the bitrate. I personally cannot hear the ringing artifiacts that these switches can produce. I consider MP3 a flawed format so going for the best-bang-for-the-buck instead of total transparency makes sense to me.
Go to the top of the page
+Quote Post
mmortal03
post Dec 5 2002, 03:17
Post #87





Group: Members
Posts: 601
Joined: 19-July 02
From: USA
Member No.: 2667



Harmonics as well as acoustics of the area (room) of the playback of the music also come into play. If you are using headphones, then it would seem to me that it would not come into play as much, but the harmonics and acoustics of original recording would MUCH MORE SO. So I guess it IS very debateable.


--------------------
WARNING: Changing of advanced parameters might degrade sound quality. Modify them only if you are expirienced in audio compression!
Go to the top of the page
+Quote Post
KikeG
post Dec 5 2002, 09:50
Post #88


WinABX developer


Group: Developer
Posts: 1578
Joined: 1-October 01
Member No.: 137



QUOTE (mithrandir @ Dec 4 2002 - 07:26 PM)
Most humans can hear over 20KHz. It's just that our ears are very insensitive in that frequency region leading us to say we can't hear that high. But if any pure signal in the 20Hz-22KHz band is strong enough, we can hear it. I'm sure most would be able to hear a 20KHz signal recorded at -5dB. However, such HF signals are almost never that loud in recorded music.

With most music, if you simply cut off all information above 16KHz, 17KHz, 18KHz, etc. it seems most listeners can't tell a single difference. That's because so much musical information is anchored in the middle frequency ranges, which subsequently help mask quiet, very high frequency information.

A fixed frequency lowpass is not a very good solution. If you set the lowpass at, say, 18.5KHz, you automatically eliminate 20KHz signals, even if they are very, very strong.

I don't think I can hear anything at 20 KHz, no matter how loud I play the test signal. Using pure tones, I can hear at reasonable volumes a 18 KHz tone. At very high volumes, I can hear a 18.5 KHz tone. And over that, I can hear nothing, no matter how loud I play the tone.

About cutoffs & real music, I've been doing some ABX tests on lowpassing using a musical clip with a very strong impulsive peak at 16 KHz (a non-isolated cymbal in a pop music song), that I've moved up and down using some resampling. I've discovered that, if concentrated, I can hear a 17.5 KHz lowpass on this clip (p<0.1%) , and seems that even a 18 KHz lowpass (p=4.9%, sequential test) buth with great pains and lots of trials. On casual listening, I doubt I could have heard the 17.5 lowpass, even the 17 KHz lowpass would have been difficult. I'd say that, in my case, a 18.5 KHz lowpass is quite safe.

I have yet to try that 'hypersonic' effect talked on the japanese paper, generating 96 KHz musical signals with some artificially recreated information over 20 KHz, and compare it with the regular version of the signal.
Go to the top of the page
+Quote Post
2Bdecided
post Dec 5 2002, 11:19
Post #89


ReplayGain developer


Group: Developer
Posts: 5066
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



QUOTE (mithrandir @ Dec 4 2002 - 06:26 PM)
Most humans can hear over 20KHz.

I think this is untrue.

I have in front of me a study from 1984 that used HF sounds up to 24 kHz, up to 124 dB (!!!). In the picture, they have what looks like a tweeter from a PA system (though I'm sure it's something much better quality) about 2 inches (!!!) from someone's ear. This is a serious attempt to try to get the subjects to detect very very loud HF sounds.

The subjects are 18-24 years old. There are 78 of them. Only 5 of them report any exposure to loud sounds (e.g. rock music, gunshots etc). Here are the number of subjects who cannot hear the maximum loudness sound of a given frequency, and the average threshold of the others.

16 kHz : 2, 47 dB
18 kHz : 3, 72 dB
20 kHz : 1, 92 dB
22 kHz : 2, 96 dB
24 kHz : 12, 100 dB

So, yes, if you try sounds MUCH LOUDER than a normal stereo can produce, many people can hear them. When they are young.

In a much older study of the same effect (where the equipment could only reach 92dB), comparable results were found - but in the older study, they looked at older listeners too.

Imagine the threshold figures I quoted above plotted on a graph, against frequency. So, as the frequency increases, the threshold curves rises steeply. As you get older, the high frequency part of the threshold graph just moves to the left. Here's the age, last frequency detected (given the 92dB limit of the equpiment), and threshold at that frequency, for three different ages:

35, 16kHz, 85dB
45, 14kHz, 83dB
55, 12kHz, 78dB

Most tweeters will melt (or burn!) before they generate 92dB at high frequencies!

So that almost proves my point, but not quite. However, in the same study, there is something even more interesting. They tested the hearing of some sudanese people, who lived in a totaly unindustrialised (i.e. quiet) rural location. They could still detect 14 kHz tones into their 70s, while the New Yorkers were loosing this ability by their 50s. Or, to put it another way, at 45, they had about 2kHz more hearing left than their New York counterparts, and this difference grew as the populations aged.

If it turns out that you can "hear" 24kHz at 150 dB, this will be academically interesting, but it won't have much impact on psychoacoustic coding. In truth, 150dB at any ultrasonic frequency will probably heat your middle ear nicely, so detection may be quite easy. But it's hardly hearing, is it?

Cheers,
David.

Henry, K. R.; and Fast, G. A. (1984).
Ultrahigh-Frequency Auditory Thresholds in Young Adults: Reliable Responses up to 24 kHz with a Quasi-Free-Field Technique.
Audiology, vol. 23, pp. 477-489.

Rosen, S.; Plester, D.; El-Mofty, A; and Rosen, H. V. (1964).
High Frequency Audiometry in Presbycusis.
Archives of Otolaryngology, vol. 79, January, pp. 18-32 or pp. 34-48.
Go to the top of the page
+Quote Post
NumLOCK
post Dec 5 2002, 11:43
Post #90


Neutrino G-RSA developer


Group: Developer
Posts: 852
Joined: 8-May 02
From: Geneva
Member No.: 2002



Very interesting survey, indeed ! Excellent documentation too.
IMHO that confirms how clever the choice of 20/22kHz for CD is.

Just a small remark on a slight inaccuracy I saw:

QUOTE
In truth, 150dB at any ultrasonic frequency will probably heat your middle ear nicely, so detection may be quite easy. But it's hardly hearing, is it?

Of course it is hearing ! wow, finally we have warm-sounding music without the disadvantages of vinyl ! laugh.gif


--------------------
Try Leeloo Chat at http://leeloo.webhop.net
Go to the top of the page
+Quote Post
Pio2001
post Dec 5 2002, 12:44
Post #91


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



QUOTE (KikeG @ Dec 5 2002 - 11:50 AM)
I have yet to try that 'hypersonic' effect talked on the japanese paper, generating 96 KHz musical signals with some artificially recreated information over 20 KHz, and compare it with the regular version of the signal.

Very difficult.
First, the information over 20 kHz was not artificial but naturally recorded at high sampling rates.
And the most tricky part : the playback did use bi-amplification to avoid intermodulation distortion under 20 kHz.

Intermodulation is very well audible, generate a 6 kHz square wave in a wave editor (at 48 kHz), perform a spectrum analysis to check that you only get 6 and 18 kHz (waveform generators can suffer from very bad alias effects), play it back in good speakers, and rise the volume for one second (not more if you don't want to fry the tweeters).
Go to the top of the page
+Quote Post
budgie
post Dec 6 2002, 09:08
Post #92





Group: Members
Posts: 341
Joined: 27-November 02
Member No.: 3901



QUOTE
Here are 5 samples. Please tell me which one is the original, and which one is the "crappy mp3" and which one is the second best... So there's only 1 original and 4 lossy audio samples.


Well, nice work from you, JohnV... Aren't you a LAME developer? tongue.gif

To start with, I must admit I never did a really true blind test... not by ABX rules. This is no apologize from me, of course. As my personal approach to music is somewhat else, I like to know what I compare. Besides, I prefer whole songs, not just approx. 20s samples sad.gif

Okay, I started with hearing all five samples just one by one. In the 2nd round I eliminated samples 2 and 4. In the 3rd round I kicked off samples 1 and 5. So the winner for me is SAMPLE THREE. The second best should be 5 or maybe also 1. Hard to tell. It all lasted about 15 minutes, I listened to it home, not in studio. I did the best because I was accused from Kblood or someone I have no guts to stand such a test... why not? It doesn't depend my life on it, neither my professional career laugh.gif
Go to the top of the page
+Quote Post
Dibrom
post Dec 6 2002, 09:31
Post #93


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE (budgie @ Dec 6 2002 - 01:08 AM)
As my personal approach to music is somewhat else, I like to know what I compare.

This is the problem with non DBT. You "like" to know what you compare, which also means you probably know what you'd "like" to win as well, and thus bias is incurred.

QUOTE
Okay, I started with hearing all five samples just one by one. In the 2nd round I eliminated samples 2 and 4. In the 3rd round I kicked off samples 1 and 5. So the winner for me is SAMPLE THREE. The second best should be 5 or maybe also 1. Hard to tell. It all lasted about 15 minutes, I listened to it home, not in studio. I did the best because I was accused from Kblood or someone I have no guts to stand such a test... why not? It doesn't depend my life on it, neither my professional career


Still no ABX results though?
Go to the top of the page
+Quote Post
floyd
post Dec 6 2002, 09:39
Post #94





Group: Members
Posts: 630
Joined: 18-June 02
Member No.: 2332



QUOTE (mithrandir @ Dec 4 2002 - 08:14 PM)
Well, that's not true. I used to be able to (and many still can) clearly hear the 19KHz signal given off by most television sets. And when I was younger I could very well hear 20KHz tones on those test discs.

So that TV tone is 19khz? I can hear that clearly, even a few rooms away (tells me someones watching the tube wink.gif ) Guess loud music hasn't totally destroyed my ears... yet.

I asked my mom if she could hear that high-pitched whine (not the best description maybe), and she looked at me like I was crazy...
Go to the top of the page
+Quote Post
NumLOCK
post Dec 6 2002, 09:50
Post #95


Neutrino G-RSA developer


Group: Developer
Posts: 852
Joined: 8-May 02
From: Geneva
Member No.: 2002



QUOTE
So that TV tone is 19khz? I can hear that clearly, even a few rooms away (tells me someones watching the tube wink.gif ) Guess loud music hasn't totally destroyed my ears... yet.

I hear it too, and it's damn annoying.. but it seems to only be present on not-too-new TV's. By the way, that 19kHz tone must be pretty much loud...

Edit: it should be forbidden to build devices that have such annoyances !

This post has been edited by NumLOCK: Dec 6 2002, 09:51


--------------------
Try Leeloo Chat at http://leeloo.webhop.net
Go to the top of the page
+Quote Post
budgie
post Dec 6 2002, 11:34
Post #96





Group: Members
Posts: 341
Joined: 27-November 02
Member No.: 3901



QUOTE
This is the problem with non DBT.  You "like" to know what you compare, which also means you probably know what you'd "like" to win as well, and thus bias is incurred.


Dear, Dibrom, for God's sake, why should I want to win?! As far as it doesn't affect someone's life or security, it really doesn't matter. I just like to compare things I really know, i.e. when I know thoroughly cons and prons of certain song/composition, I can much easier compare then, because I just know exactly, where to be extremely cautious... that's all. It helps me a lot. Besides, I am convinced that cutoffs at high frequencies really somewhat affect what we hear, even if we are not able to explain it exactly. But the resulting impression is somewhat affected. And I am not trained in such way, to compare short samples, I would rather prefer all the song. And what makes me "nervous" tongue.gif is the fact, that I do not know, what I hear laugh.gif laugh.gif laugh.gif
Go to the top of the page
+Quote Post
JohnV
post Dec 6 2002, 11:52
Post #97





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



QUOTE (budgie @ Dec 6 2002 - 10:08 AM)
Well, nice work from you, JohnV... Aren't you a LAME developer?  :P

To start with, I must admit I never did a really true blind test... not by ABX rules. This is no apologize from me, of course. As my personal approach to music is somewhat else, I like to know what I compare. Besides, I prefer whole songs, not just approx. 20s samples  :(

Okay, I started with hearing all five samples just one by one. In the 2nd round I eliminated samples 2 and 4. In the 3rd round I kicked off samples 1 and 5. So the winner for me is SAMPLE THREE. The second best should be 5 or maybe also 1. Hard to tell. It all lasted about 15 minutes, I listened to it home, not in studio. I did the best because I was accused from Kblood or someone I have no guts to stand such a test... why not? It doesn't depend my life on it, neither my professional career  :lol:

I'm not a LAME developer, although I like to help Lame development.. wink.gif

the test files are:
Wait1: Psytel AAC 2.15 -archive -disable_ms
Wait2: Lame MP3 3.90.2 --alt-preset insane -k
Wait3: MPC Mppenc 1.01j --insane --minSMR 3
Wait4: Original
Wait5 Ogg Vorbis oggenc 1.0 -b320

Such settings were used which give lots of high freqs, and I did to all samples (including the original) a 20khz lowpass in CEP2pro, in order to make the spectral recognition not TOTALLY obvious.. Yes, I tried to make spectral recognition harder, so the files are not optimal quality..at least MP3 and MPC..

This post has been edited by JohnV: Dec 6 2002, 12:07


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
JohnV
post Dec 6 2002, 12:06
Post #98





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



QUOTE (floyd @ Dec 6 2002 - 10:39 AM)
So that TV tone is 19khz?  I can hear that clearly, even a few rooms away (tells me someones watching the tube wink.gif )  Guess loud music hasn't totally destroyed my ears... yet.

Eeh, wasn't it something like 16kHz? Also tv-sets can give many kind of high pitched sounds.. I hear different hf sounds pretty much from every tv.. laugh.gif


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
budgie
post Dec 6 2002, 12:10
Post #99





Group: Members
Posts: 341
Joined: 27-November 02
Member No.: 3901



QUOTE (JohnV @ Dec 6 2002 - 02:52 AM)
QUOTE (budgie @ Dec 6 2002 - 10:08 AM)
Well, nice work from you, JohnV... Aren't you a LAME developer?  :P

To start with, I must admit I never did a really true blind test... not by ABX rules. This is no apologize from me, of course. As my personal approach to music is somewhat else, I like to know what I compare. Besides, I prefer whole songs, not just approx. 20s samples  :(

Okay, I started with hearing all five samples just one by one. In the 2nd round I eliminated samples 2 and 4. In the 3rd round I kicked off samples 1 and 5. So the winner for me is SAMPLE THREE. The second best should be 5 or maybe also 1. Hard to tell. It all lasted about 15 minutes, I listened to it home, not in studio. I did the best because I was accused from Kblood or someone I have no guts to stand such a test... why not? It doesn't depend my life on it, neither my professional career  :lol:

I'm not a LAME developer, although I like to help Lame development.. wink.gif

the test files are:
Wait1: Psytel AAC 2.15 -archive -disable_ms
Wait2: Lame MP3 3.90.2 --alt-preset insane -k
Wait3: MPC Mppenc 1.01j --insane --minSMR 3
Wait4: Original
Wait5 Ogg Vorbis oggenc 1.0 -b320

Such settings were used which give lots of high freqs, and I did to all samples (including the original) a 20khz lowpass in CEP2pro, in order to make the spectral recognition not TOTALLY obvious.. Yes, I tried to make spectral recognition harder, so the files are not optimal quality..

QUOTE
Such settings were used which give lots of high freqs, and I did to all samples (including the original) a 20khz lowpass in CEP2pro, in order to make the spectral recognition not TOTALLY obvious.. Yes, I tried to make spectral recognition harder, so the files are not optimal quality.


Well, I lost. Sad, but true mad.gif I am not very pleased by the fact, indeed.

But what I appreciate, that you admit, YOU WERE CHEATING! Because it is not ORIGINAL ANYMORE. Just read your words carefully again. laugh.gif You promised me 1 ORIGINAL!!! But what you sent to me, was no original wave... sad.gif
Go to the top of the page
+Quote Post
Volcano
post Dec 6 2002, 12:16
Post #100





Group: Members (Donating)
Posts: 916
Joined: 30-September 01
From: Berlin, Germany
Member No.: 112



JohnV:

*argh*!

Why did you have to give out the results, I would have loved to see some ABX results. laugh.gif


bugdie:

We've heard excuses like yours (need a complete song for comparison, etc.) a million times before. rolleyes.gif The fact that the sample was slightly manipulated doesn't make any difference.
Go to the top of the page
+Quote Post

8 Pages V  « < 2 3 4 5 6 > » 
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: 1st August 2014 - 01:57