Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lame 3.97b1 vs Original CD (Read 17168 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lame 3.97b1 vs Original CD

Reply #25 –
Quote
Oh Lyx, Did you try an abx between different LAME presets? If you did what were your results? Can you try Insane vs fast standard? That would be more interesting cause after doing the wav to mp3 test, I couldn't tell the difference between insane and fast standard with this sample. Maybe I should try it again before any fatigue
[a href="index.php?act=findpost&pid=330807"][{POST_SNAPBACK}][/a]

Nope, i didn't and am not gonna do that. My hearing-abilities in some frequency-ranges and soundtypes is a problem to me - to such an extend that i got minor headaches when listening to early versions of vorbis(in mid-to-high bitrates, this problem has mostly been solved in current vorbis-versions). Therefore, i avoid taking part in codec-tests because i dont want to train myself in becoming even "better" in that regard.

My intention when doing the quick test above was not to pinpoint the problem, but instead just to verify your findings that at high bitrate-ranges, a problem actually exists with this sample(because until then, your findings were unverified in the thread), while putting the least necessary amount of stress on my ears(therefore the APFM-encoding).

Some hints about how to easier find differences:
- try testing after you are just 2hours awake, and before you are exposed to "normal" loudness-levels.
- "recalibrate" yourself to the originals(A and B) every 2-3 trials.
- try to find a single spot in the sample where the problem is very obvious to you and get "familiar" with that single spot.

You may especially want to try if the highest VBR-preset (V0) mostly solves the problem, and then in the future encode this kind of audiomaterial at V0.

- Lyx
I am arrogant and I can afford it because I deliver.

Lame 3.97b1 vs Original CD

Reply #26 –
Quote
Quote
Oh Lyx, Did you try an abx between different LAME presets? If you did what were your results? Can you try Insane vs fast standard? That would be more interesting cause after doing the wav to mp3 test, I couldn't tell the difference between insane and fast standard with this sample. Maybe I should try it again before any fatigue
[a href="index.php?act=findpost&pid=330807"][{POST_SNAPBACK}][/a]

Nope, i didn't and am not gonna do that. My hearing-abilities in some frequency-ranges and soundtypes is a problem to me - to such an extend that i got minor headaches when listening to early versions of vorbis(in mid-to-high bitrates, this problem has mostly been solved in current vorbis-versions). Therefore, i avoid taking part in codec-tests because i dont want to train myself in becoming even "better" in that regard.

My intention when doing the quick test above was not to pinpoint the problem, but instead just to verify your findings that at high bitrate-ranges, a problem actually exists with this sample(because until then, your findings were unverified in the thread), while putting the least necessary amount of stress on my ears(therefore the APFM-encoding).

Some hints about how to easier find differences:
- try testing after you are just 2hours awake, and before you are exposed to "normal" loudness-levels.
- "recalibrate" yourself to the originals(A and B) every 2-3 trials.
- try to find a single spot in the sample where the problem is very obvious to you and get "familiar" with that single spot.

You may especially want to try if the highest VBR-preset (V0) mostly solves the problem, and then in the future encode this kind of audiomaterial at V0.

- Lyx
[a href="index.php?act=findpost&pid=330808"][{POST_SNAPBACK}][/a]



It is actually stressful because i keep on recalibrating my ears to the specific part of sample A, do it many times on B. Switch back and forth, then I listen to x and y once then instantly make my decision.

Lame 3.97b1 vs Original CD

Reply #27 –
Does anyone else have any ABX test results? preferably fast standard vs insane.

Lame 3.97b1 vs Original CD

Reply #28 –
I never start a (useless) comparison between -V2 & CBR 320, but I have several samples which sound better with CBR 320.
-V2 is supposed to be very good, transparent for most people and on most samples, but it's not a miraculous command line. Distortions are therefore still possible. CBR 320 would be a perfect choice... if bitrate wasn't so high. -V2 (you can also try -V3) is most often as good but at much lower bitrate (180...220). In other word, -V2 is more efficient (but never better than CBR 320).
Wavpack Hybrid: one encoder for all scenarios
WavPack -c4.5hx6 (44100Hz & 48000Hz) ≈ 390 kbps + correction file
WavPack -c4hx6 (96000Hz) ≈ 768 kbps + correction file
WavPack -h (SACD & DSD) ≈ 2400 kbps at 2.8224 MHz

Lame 3.97b1 vs Original CD

Reply #29 –
Quote
I never start a (useless) comparison between -V2 & CBR 320, but I have several samples which sound better with CBR 320.
-V2 is supposed to be very good, transparent for most people and on most samples, but it's not a miraculous command line. Distortions are therefore still possible. CBR 320 would be a perfect choice... if bitrate wasn't so high. -V2 (you can also try -V3) is most often as good but at much lower bitrate (180...220). In other word, -V2 is more efficient (but never better than CBR 320).
[a href="index.php?act=findpost&pid=330815"][{POST_SNAPBACK}][/a]


I know 320 will always be better than preset standard but the whole reason I made this post was to say the supposed best mp3 format is still distinguishable from wav and not transparent. Now, Since preset fast standard is supposed to be the minimum transparent setting, It would be interesting to know if people could still distinguish from insane and standard setting for this sample. If people can't distinguish between the two, this will just confirm that preset standard is a very good sound/size ratio and that going 320 is not going to approach the sound of the original WAV file much more

Lame 3.97b1 vs Original CD

Reply #30 –
Oh by the way guruboolez, Where can I access these files that sound better with 320 than standard? do you have them posted somewhere that I can download them? Thanks

Lame 3.97b1 vs Original CD

Reply #31 –
Quote
I made this post was to say the supposed best mp3 format is still distinguishable from wav and not transparent.

If this was really the reason, then i'm sorry, you didn't prove anything new. PStandard was never ever considered to be transparent on every sample in the world to every listener in the world. Not even 320kbit is transparent by those criteria.

What PStandard claims, that it is indistinguishable to the original in *almost* every case even for trained listeners. So, like "more than good enough" instead of "perfect". The fact that even your sample which is an extreme example (microattacks) could not be verified by 2 people, and required close attention from someone who is very sensible to HF-transients, only shows that PStandard is indeed transparent, except in rare cases.... and even in those rare cases, the result is not that it sounds bad, but instead just "okay" instead of "perfect".
I am arrogant and I can afford it because I deliver.

Lame 3.97b1 vs Original CD

Reply #32 –
Quote
Quote
I made this post was to say the supposed best mp3 format is still distinguishable from wav and not transparent.

If this was really the reason, then i'm sorry, you didn't prove anything new. PStandard was never ever considered to be transparent on every sample in the world to every listener in the world. Not even 320kbit is transparent by those criteria.

What PStandard claims, that it is indistinguishable to the original in *almost* every case even for trained listeners. So, like "more than good enough" instead of "perfect". The fact that even your sample which is an extreme example (microattacks) could not be verified by 2 people, and required close attention from someone who is very sensible to HF-transients, only shows that PStandard is indeed transparent, except in rare cases.... and even in those rare cases, the result is not that it sounds bad, but instead just "okay" instead of "perfect".
[a href="index.php?act=findpost&pid=330848"][{POST_SNAPBACK}][/a]


So the question now is... Is there a lossy format that is truly transparent to all peoples ears? What is it and what bitrate does it sound transparent to all known testers?

Lame 3.97b1 vs Original CD

Reply #33 –
Quote
So the question now is... Is there a lossy format that is truly transparent to all peoples ears? What is it and what bitrate does it sound transparent to all known testers?

I doubt that a "100% perfect" lossy-format exists - at reasonable bitrate-ranges (400kbit would be quite stupid - might as well go lossless then), but MPC at the insane-preset might be a candidate. Vorbis overally is considered on-par with MPC, but with this kind of audiomaterial, i doubt that it performs better than MPC, because this sample almost asks for 2 vorbis-vulnerabilities at once: pre-echo and HF-transients simultaneusly(*if* vorbis fails on this sample, then the effect will probably be the opposite of mp3 - the short HF-attacks will sound more pronounced and clear instead of less). The main reason why i mention MPC because it is the format with the lowest amount of "problem-cases", thus it is a very "robust" format.
I am arrogant and I can afford it because I deliver.

Lame 3.97b1 vs Original CD

Reply #34 –
Quote
Quote
Quote
I made this post was to say the supposed best mp3 format is still distinguishable from wav and not transparent.

If this was really the reason, then i'm sorry, you didn't prove anything new. PStandard was never ever considered to be transparent on every sample in the world to every listener in the world. Not even 320kbit is transparent by those criteria.

What PStandard claims, that it is indistinguishable to the original in *almost* every case even for trained listeners. So, like "more than good enough" instead of "perfect". The fact that even your sample which is an extreme example (microattacks) could not be verified by 2 people, and required close attention from someone who is very sensible to HF-transients, only shows that PStandard is indeed transparent, except in rare cases.... and even in those rare cases, the result is not that it sounds bad, but instead just "okay" instead of "perfect".
[a href="index.php?act=findpost&pid=330848"][{POST_SNAPBACK}][/a]


So the question now is... Is there a lossy format that is truly transparent to all peoples ears? What is it and what bitrate does it sound transparent to all known testers?
[a href="index.php?act=findpost&pid=330861"][{POST_SNAPBACK}][/a]


lossy means with loss, so eventually someone will be able to spot a difference against the original.
Short answer to your question: no

Lame 3.97b1 vs Original CD

Reply #35 –
Quote
Quote
Quote
Quote
I made this post was to say the supposed best mp3 format is still distinguishable from wav and not transparent.

If this was really the reason, then i'm sorry, you didn't prove anything new. PStandard was never ever considered to be transparent on every sample in the world to every listener in the world. Not even 320kbit is transparent by those criteria.

What PStandard claims, that it is indistinguishable to the original in *almost* every case even for trained listeners. So, like "more than good enough" instead of "perfect". The fact that even your sample which is an extreme example (microattacks) could not be verified by 2 people, and required close attention from someone who is very sensible to HF-transients, only shows that PStandard is indeed transparent, except in rare cases.... and even in those rare cases, the result is not that it sounds bad, but instead just "okay" instead of "perfect".
[a href="index.php?act=findpost&pid=330848"][{POST_SNAPBACK}][/a]


So the question now is... Is there a lossy format that is truly transparent to all peoples ears? What is it and what bitrate does it sound transparent to all known testers?
[a href="index.php?act=findpost&pid=330861"][{POST_SNAPBACK}][/a]


lossy means with loss, so eventually someone will be able to spot a difference against the original.
Short answer to your question: no
[a href="index.php?act=findpost&pid=330866"][{POST_SNAPBACK}][/a]


Well lossy for encoding means the original data cannot be reconstructed unlike lossless which just acts like a zip file. This doesn't mean that it cannot produce exactly the same sound as the original if implemented perfectly. TOTAL silence for example can be reproduced with a bitrate of 1 if it exists, therefore it only takes this amount of data to make it sound like the original uncompressed file. I'm sure one day the developers will get it just right and make a perfectly efficient lossy encoder. Well, even the preset fast standard mp3 really isn't that far off considering its about 7x smaller than the original wav.

Maybe one day the only reason one might stay with lossless is to be able to convert back and forth among different formats without any generation loss

Lame 3.97b1 vs Original CD

Reply #36 –
Quote
I'm sure one day the developers will get it just right and make a perfectly efficient lossy encoder.

Far earlier, storage-capacities and bandwidth will have risen enough to make lossy so unnecessary that it becomes a niche-product, therefore never reaching the "holy-grail" - because it was never really needed - it was a "good enough"-approach from the beginning on.

Quote
Well, even the preset fast standard mp3 really isn't that far off considering its about 7x smaller than the original wav.

Right, and if this is the case for 99,999% listener/sample combos in the world(agreed, the number could be lower if more people are trained to detect artifacts... but that wont happen for obvious reasons), then why spent ten-thousands of hours into fixing the remaining 0,001%, especially when its to expected that it will be rendered useless by lossless becoming the norm in the future anyways? That effort/gain-ratio just doesnt make sense.

Quote
Maybe one day the only reason one might stay with lossless is to be able to convert back and forth among different formats without any generation loss

In am optimal world, and if humans would think in resource-saving aspects, you could be right. But unfortunatelly the opposite is the case: any additional resources get saturated immediatelly(therefore requiring ever more resources).
I am arrogant and I can afford it because I deliver.

Lame 3.97b1 vs Original CD

Reply #37 –
Quote
lossy means with loss, so eventually someone will be able to spot a difference against the original.
Short answer to your question: no

Quote

Well lossy for encoding means the original data cannot be reconstructed unlike lossless which just acts like a zip file. This doesn't mean that it cannot produce exactly the same sound as the original if implemented perfectly. TOTAL silence for example can be reproduced with a bitrate of 1 if it exists, therefore it only takes this amount of data to make it sound like the original uncompressed file. I'm sure one day the developers will get it just right and make a perfectly efficient lossy encoder. Well, even the preset fast standard mp3 really isn't that far off considering its about 7x smaller than the original wav.

Maybe one day the only reason one might stay with lossless is to be able to convert back and forth among different formats without any generation loss
[a href="index.php?act=findpost&pid=330871"][{POST_SNAPBACK}][/a]


There is no such thing as perfect implementation. There is always room for mistakes. And you cannot cover all the statistical universe.
Another thingy: one thing is coming from 1411kbps to 200kbps (7x compression) and keeping perceived quality. Another thing is coming from 200kbps to 32kbps (another 7x compression from 200kbps). Things are not that linear.

Lame 3.97b1 vs Original CD

Reply #38 –
Quote
Quote
I'm sure one day the developers will get it just right and make a perfectly efficient lossy encoder.

Far earlier, storage-capacities and bandwidth will have risen enough to make lossy so unnecessary that it becomes a niche-product, therefore never reaching the "holy-grail" - because it was never really needed - it was a "good enough"-approach from the beginning on.

Quote
Well, even the preset fast standard mp3 really isn't that far off considering its about 7x smaller than the original wav.

Right, and if this is the case for 99,999% listener/sample combos in the world(agreed, the number could be lower if more people are trained to detect artifacts... but that wont happen for obvious reasons), then why spent ten-thousands of hours into fixing the remaining 0,001%, especially when its to expected that it will be rendered useless by lossless becoming the norm in the future anyways? That effort/gain-ratio just doesnt make sense.

Quote
Maybe one day the only reason one might stay with lossless is to be able to convert back and forth among different formats without any generation loss

In am optimal world, and if humans would think in resource-saving aspects, you could be right. But unfortunatelly the opposite is the case: any additional resources get saturated immediatelly(therefore requiring ever more resources).
[a href="index.php?act=findpost&pid=330873"][{POST_SNAPBACK}][/a]


I guess you're right. The only compromise is that there just isn't a DAP existing today that can hold 3000 albums worth of lossless files which could be useful if you want to stay mobile and are still very quality conscious. Oh well