MP3 encoder sounds terrible

Page: < 12 Showing page 2 of 2
Author
rbowser
Max Output Level: -10 dBFS
  • Total Posts : 6518
  • Joined: 2005/07/31 14:32:34
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 16:29:58 (permalink)
rosstracy


Coming in a bit late, but I initially had issues with the encoder from Sonar and found a few things made a difference:

1) My mixing left a little to be desired, I rolled back and redid all my EQs because of bad monitor/room setup.
2) I bumped up to 192 from 128, and it made a huge difference.  Not sure if I'd go to 320 for an MP3, but I may try it.
3) The idea above of a 2-track master WAV is probably a good way to go, minimally for process flow and efficiency, and then go to MP3.

Essentially, the mix and 192 bit rate made the big difference for me.


Yes, this idea of only having an MP3 to show for one's work didn't use to come up, but it does more often now since everyone uses MP3s daily with their portable devices.  But to have only an MP3 of a mix is akin to only keeping a photocopy of an original document - ya know?

The 128 bit rate is impossible.  Ever listen to things at SoundClick?- those are 128, and that's why things sound so lousy there.

Randy B.

Sonar X3e Studio
Roland A-800 MIDI keyboard controller
Alesis i|O2 interface
Gigabyte Technology-AMD Phenom II @ 3 GHz
8 Gb RAM 6 Core Windows 7 Home Premium x64
with dual monitors
#31
UnderTow
Max Output Level: -37 dBFS
  • Total Posts : 3848
  • Joined: 2004/01/06 12:13:49
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 17:33:09 (permalink)
BEATZM1D10T


That's not entirely true. 192Khz picks up frequencies well above the human hearing spectrum that manifest themselves in sub-harmonics within the audible range of humans. 44.1Khz does not pick up those frequencies. But, that's for another discussion for another time.
This is not correct but indeed, it is another discussion for another time.

UnderTow
#32
UnderTow
Max Output Level: -37 dBFS
  • Total Posts : 3848
  • Joined: 2004/01/06 12:13:49
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 17:38:06 (permalink)
WDI

If the 192 MP3 still sounds like garbage there most likely is another problem.
Oh absolutely. I am just a stickler for technical details. :-)





UnderTow


#33
BEATZM1D10T
Max Output Level: -85 dBFS
  • Total Posts : 285
  • Joined: 2009/05/22 12:43:50
  • Location: Mid-West
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 18:39:15 (permalink)
UnderTow


BEATZM1D10T


That's not entirely true. 192Khz picks up frequencies well above the human hearing spectrum that manifest themselves in sub-harmonics within the audible range of humans. 44.1Khz does not pick up those frequencies. But, that's for another discussion for another time.
This is not correct but indeed, it is another discussion for another time.

UnderTow
/hijack

http://www.aes.org/e-lib/browse.cfm?elib=15398 - "Overall, participants were able to discriminate between files recorded at 88.2kHz and their 44.1kHz down-sampled version. Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2kHz and files recorded at 44.1kHz."

[link=http://asadl.org/jasa/resource/1/jasman/v117/i4/p2147_s1?isAuthorized=no]http://asadl.org/jasa/res...147_s1?isAuthorized=no
[/link] - basically HF information helps us differentiate space.

http://en.wikipedia.org/wiki/Nyquist_frequency

http://www.searsound.com/philosophy.html - Read 'The Recorded Sound Sucks - We're trying to make it better'. Written by Walter Sear. All of his articles are great to read honestly for someone with an interest in our craft.

I have colleagues, who went to highly esteemed audio engineering schools, who's instructors could recognize what sample rate projects were recorded in a heart beat.

Just because you can't hear a difference between a 44.1khz/192khz recording (and your outboard gear's ability to reproduce this response) and their impact on the recorded product, doesn't mean there isn't a perceivable difference to those of us that can.

I'll side with the AES and established (highly regarded) professionals in our field on this one.

/endhijack
post edited by BEATZM1D10T - 2011/04/14 18:42:39
#34
SteveXN
Max Output Level: -89 dBFS
  • Total Posts : 63
  • Joined: 2007/12/01 12:11:22
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 18:40:05 (permalink)
Thanks ALL! 31 comments in 23 hours. Must be a boring day. first, I changed bit rate to 192 and that made all the difference. Problem fixed. Here's where steve shows his ignorance: I know how to use export to wav, but I've never "bounced" anything and have no idea what it means or how to create a stereo master wav other than by exporting to a file. If this latter is what you mean, then I don't know how to select an exported wave file with the mp3 converter from within Sonar and I don't know how to access it outside of Sonar. Next, I never select any tracks for export, I intentionally mouse click outside the track area so that no tracks are selected and that has always worked. Next, for this tune I was trying to make a $75 upright bass sound like a $10,000 one and eq'd the crap out of it, creating some of the issues no doubt. Next, I've been trying to use the Boost 11 and have gone from +6db down to +3db and am considering aborting its use, knowing nothing about how it should be used. I blame sonar for pricing their software so reasonably with so many features that anyone with no experience or time to invest can pretend to be a recording engineer :D  Anyone interested in enlightening me on any of the above; I'm all ears. Thanks again

These are the works of man? This is the sum of our ambition? Sting

Sonar Platinum, Sapphire Pro40, Gateway DX4840, i3 550 @ 3.20GHz, 64 bit, 8GBRam, Shure KSM27/SM81, Takamine
#35
SteveXN
Max Output Level: -89 dBFS
  • Total Posts : 63
  • Joined: 2007/12/01 12:11:22
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 18:59:09 (permalink)
If anyone wants to hear the 192 bit version it is here. I was going to do some vocal retakes, but it may be interesting to you as is

http://www.soundclick.com...&q=hi&newref=1


These are the works of man? This is the sum of our ambition? Sting

Sonar Platinum, Sapphire Pro40, Gateway DX4840, i3 550 @ 3.20GHz, 64 bit, 8GBRam, Shure KSM27/SM81, Takamine
#36
BEATZM1D10T
Max Output Level: -85 dBFS
  • Total Posts : 285
  • Joined: 2009/05/22 12:43:50
  • Location: Mid-West
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 19:05:23 (permalink)
SteveXN


If anyone wants to hear the 192 bit version it is here. I was going to do some vocal retakes, but it may be interesting to you as is

http://www.soundclick.com...&q=hi&newref=1



Very nice, I like the style.
#37
chuckebaby
Max Output Level: 0 dBFS
  • Total Posts : 13146
  • Joined: 2011/01/04 14:55:28
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/14 19:24:51 (permalink)
BEATZM1D10T


UnderTow


BEATZM1D10T


That's not entirely true. 192Khz picks up frequencies well above the human hearing spectrum that manifest themselves in sub-harmonics within the audible range of humans. 44.1Khz does not pick up those frequencies. But, that's for another discussion for another time.
This is not correct but indeed, it is another discussion for another time.

UnderTow
/hijack

http://www.aes.org/e-lib/browse.cfm?elib=15398%3C/a%3E - "Overall, participants were able to discriminate between files recorded at 88.2kHz and their 44.1kHz down-sampled version. Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2kHz and files recorded at 44.1kHz."

[link=http://asadl.org/jasa/resource/1/jasman/v117/i4/p2147_s1?isAuthorized=no]http://asadl.org/jasa/res...147_s1?isAuthorized=no
[/link] - basically HF information helps us differentiate space.

http://en.wikipedia.org/wiki/Nyquist_frequency

http://www.searsound.com/philosophy.html - Read 'The Recorded Sound Sucks - We're trying to make it better'. Written by Walter Sear. All of his articles are great to read honestly for someone with an interest in our craft.

I have colleagues, who went to highly esteemed audio engineering schools, who's instructors could recognize what sample rate projects were recorded in a heart beat.

Just because you can't hear a difference between a 44.1khz/192khz recording (and your outboard gear's ability to reproduce this response) and their impact on the recorded product, doesn't mean there isn't a perceivable difference to those of us that can.

I'll side with the AES and established (highly regarded) professionals in our field on this one.

/endhijack


the man is packing amomo.

Windows 8.1 X64 Sonar Platinum x64
Custom built: Asrock z97 1150 - Intel I7 4790k - 16GB corsair DDR3 1600 - PNY SSD 220GB
Focusrite Saffire 18I8 - Mackie Control
   
#38
tarsier
Max Output Level: -45 dBFS
  • Total Posts : 3029
  • Joined: 2003/11/07 11:51:35
  • Location: 6 feet under
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 10:43:31 (permalink)
Thanks for the reply, Tarsier - When I make an MP3 with Sonar, I use Lame Encoder v. 3.97. Outside of Sonar, I use Sony MP3 v.3. In both cases, I only make MP3s from 2-track .wav masters, never directly from a project's mix. My Windows Media Player is version 11.0.6. I prefer using VLC because I've never found a file type it can't play. It's version 1.0.3

There are a lot of variables there.  Does your VBR problem happen with all of those encoders and playback apps?
#39
UnderTow
Max Output Level: -37 dBFS
  • Total Posts : 3848
  • Joined: 2004/01/06 12:13:49
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 10:48:01 (permalink)

chuckebaby

the man is packing amomo.
The man is packing blanks which are misfiring while he is shooting randomly in all directions in the hope of hitting the target he can't seem to see.

BEATZM1D10T



UnderTow


BEATZM1D10T


That's not entirely true. 192Khz picks up frequencies well above the human hearing spectrum that manifest themselves in sub-harmonics within the audible range of humans. 44.1Khz does not pick up those frequencies. But, that's for another discussion for another time.
This is not correct but indeed, it is another discussion for another time.

UnderTow
/hijack

http://www.aes.org/e-lib/browse.cfm?elib=15398 - "Overall, participants were able to discriminate between files recorded at 88.2kHz and their 44.1kHz down-sampled version. Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2kHz and files recorded at 44.1kHz."
That research has been debunked. Not only is the test itself flawed but the conclusion that you quote does not even follow from the test results!

1. The recordings at the different sample rates are not made with the same clocks. This in itself already invalidates the test but let's continue.

2. To exclude issues with the playback sample rate, the 44.1 Khz recording and down-sampled versions should be resampled to 88.2Khz so that all samples can be auditioned at that rate. This was not done.

3. The re-sampling was performed with the internal Pyramix converter. This is what a Sine Sweep looks like after re-sampling with Pyramix (in Version 6 which was used for this test):



This is the same Sine Sweep converted with the excellent and free SoX converter:



(Note that Pyramix is quite popular in the "Hi-Rez"circles. Makes you wonder heh? ;-) )

4. No clear description of how the ABX testing was performed.

5. The test results: Of the 16 participants 13 did not perform better than chance and 3 seem to have performed better than chance but in the wrong way: They got the answer wrong a "statistically significant" number of times!

6. The next issue with the results is that instead of deciding what is relevant before the tests start and then look at the probability of the results occurring by chance, it seems they calculated the probability of every possible permutation of each individual listener and each individual chance (346 p-values) and then cherry picked the 12 "significant" ones. The thing they seem to be missing is that with so many values there will be a number of false positives by pure chance! That is the nature of statistics. The three candidates that had "significant results" were chosen for these p-values, after the fact! They were not chosen beforehand because of a background in audio engineering or something like that. This is fishy at best.


[link=http://asadl.org/jasa/resource/1/jasman/v117/i4/p2147_s1?isAuthorized=no]http://asadl.org/jasa/res...147_s1?isAuthorized=no[/link] - basically HF information helps us differentiate space.
There is nothing in the resumé of that article that says anything about frequencies above 20Khz (or any mention of what frequencies were tested actually). I am not going to pay 20$ to buy the article for the sole purpose of proving you wrong. ;-)  Not least because much of this study is based on simulations. The article isn't even about the whole auditory system. Just a tine subsection of it. 
http://en.wikipedia.org/wiki/Nyquist_frequency
Yes I am fully aware of the Nyquist Frequency. What is your point? If you are referring to this particular sentence "However, this reconstruction requires an ideal filter that passes some frequencies unchanged while suppressing all others completely (commonly called a brick-wall filter). In practice, perfect reconstruction is unattainable. Some amount of aliasing is unavoidable." then note that there is no mention of audible aliasing. That is a crucial point. For instance if you use SoX to re-sample, artefacts are 170 dB below the signal. It doesn't matter how good you hear or how good your monitoring is, there is no way whatsoever anyone can hear that.

http://www.searsound.com/philosophy.html - Read 'The Recorded Sound Sucks - We're trying to make it better'. Written by Walter Sear. All of his articles are great to read honestly for someone with an interest in our craft.
This is only opinion with zero science to support it. That interview actually very clearly shows he did not understand sampling theory. He says "The bandpass, the frequency response - much, much too limited. It will improve as they go up to 192. The follow the dots becomes less follow the dots. There are more dots." and "and until the dots get close enough so that the ear can't hear the difference. It's like the old dot matrix printers - you could tell it was an "O" with the little steps up and the little steps down - you could read it as an "O" - but it wasn't very elegant." That was utterly clueless. The guy had no idea how sampling works.
I have colleagues, who went to highly esteemed audio engineering schools, who's instructors could recognize what sample rate projects were recorded in a heart beat.
More anecdotal evidence. It is meaningless. I have heard many many people claim they can hear differences. Once put to the test, they all fail. (And not just with me. With any proper test ever performed anywhere).
Just because you can't hear a difference between a 44.1khz/192khz recording (and your outboard gear's ability to reproduce this response) and their impact on the recorded product, doesn't mean there isn't a perceivable difference to those of us that can.
Ah yes that childish argument. Question my hearing. I could also stoop to that level and call you delusional but what is the point? All science and all evidence points to higher sampling rates being indistinguishable from 44.1Khz. On the other hand expectation bias and placebo are something every single person on the planet is subject to. There is no question about this. 99% of science is about procedures to avoid and work around the flawed human bias. Anyone that claims different needs to come with some very serious and solid science based evidence. Extraordinary claims need extraordinary proof.

On my quality converters, (Prism Sound), no one can tell the difference between different sample rates in a double blind ABX test. That includes many people that told me they could easily tell the difference. Some claimed night and day differences or other such superlatives... that is usually a clear give away that either their converters are completely broken or they have an over active imagination.

I'll side with the AES and established (highly regarded) professionals in our field on this one.
I'll side with A) the scientists and B) the actual proof and science. But you know what what? Most papers at the AES agree with the science and the proof.  (Note that having done a presentation at the AES does not mean that the AES agrees with your conclusions). I don't care how many grammys someone has, their claims are meaningless unless they have solid science and real testing to back-up them up. It is relatively easy to setup a proper double-blind test to demonstrate this yet so far no one, absolutely no one, has managed to. I think the conclusions are more than obvious.

UnderTow
post edited by UnderTow - 2011/04/15 11:06:37
#40
The Maillard Reaction
Max Output Level: 0 dBFS
  • Total Posts : 31918
  • Joined: 2004/07/09 20:02:20
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 10:54:11 (permalink)

Highly esteemed audio engineering schools?




#41
chuckebaby
Max Output Level: 0 dBFS
  • Total Posts : 13146
  • Joined: 2011/01/04 14:55:28
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 11:24:12 (permalink)
UnderTow



chuckebaby

the man is packing amomo.
The man is packing blanks which are misfiring while he is shooting randomly in all directions in the hope of hitting the target he can't seem to see.

BEATZM1D10T



UnderTow


BEATZM1D10T


That's not entirely true. 192Khz picks up frequencies well above the human hearing spectrum that manifest themselves in sub-harmonics within the audible range of humans. 44.1Khz does not pick up those frequencies. But, that's for another discussion for another time.
This is not correct but indeed, it is another discussion for another time.

UnderTow
/hijack

http://www.aes.org/e-lib/browse.cfm?elib=15398%3C/a%3E - "Overall, participants were able to discriminate between files recorded at 88.2kHz and their 44.1kHz down-sampled version. Furthermore, for the orchestral excerpt, they were able to discriminate between files recorded at 88.2kHz and files recorded at 44.1kHz."
That research has been debunked. Not only is the test itself flawed but the conclusion that you quote does not even follow from the test results!

1. The recordings at the different sample rates are not made with the same clocks. This in itself already invalidates the test but let's continue.

2. To exclude issues with the playback sample rate, the 44.1 Khz recording and down-sampled versions should be resampled to 88.2Khz so that all samples can be auditioned at that rate. This was not done.

3. The re-sampling was performed with the internal Pyramix converter. This is what a Sine Sweep looks like after re-sampling with Pyramix (in Version 6 which was used for this test):



This is the same Sine Sweep converted with the excellent and free SoX converter:



(Note that Pyramix is quite popular in the "Hi-Rez"circles. Makes you wonder heh? ;-) )

4. No clear description of how the ABX testing was performed.

5. The test results: Of the 16 participants 13 did not perform better than chance and 3 seem to have performed better than chance but in the wrong way: They got the answer wrong a "statistically significant" number of times!

6. The next issue with the results is that instead of deciding what is relevant before the tests start and then look at the probability of the results occurring by chance, it seems they calculated the probability of every possible permutation of each individual listener and each individual chance (346 p-values) and then cherry picked the 12 "significant" ones. The thing they seem to be missing is that with so many values there will be a number of false positives by pure chance! That is the nature of statistics. The three candidates that had "significant results" were chosen for these p-values, after the fact! They were not chosen beforehand because of a background in audio engineering or something like that. This is fishy at best.


[link=http://asadl.org/jasa/resource/1/jasman/v117/i4/p2147_s1?isAuthorized=no]http://asadl.org/jasa/res...147_s1?isAuthorized=no[/link] - basically HF information helps us differentiate space.
There is nothing in the resumé of that article that says anything about frequencies above 20Khz (or any mention of what frequencies were tested actually). I am not going to pay 20$ to buy the article for the sole purpose of proving you wrong. ;-)  Not least because much of this study is based on simulations. The article isn't even about the whole auditory system. Just a tine subsection of it. 
http://en.wikipedia.org/wiki/Nyquist_frequency
Yes I am fully aware of the Nyquist Frequency. What is your point? If you are referring to this particular sentence "However, this reconstruction requires an ideal filter that passes some frequencies unchanged while suppressing all others completely (commonly called a brick-wall filter). In practice, perfect reconstruction is unattainable. Some amount of aliasing is unavoidable." then note that there is no mention of audible aliasing. That is a crucial point. For instance if you use SoX to re-sample, artefacts are 170 dB below the signal. It doesn't matter how good you hear or how good your monitoring is, there is no way whatsoever anyone can hear that.

http://www.searsound.com/philosophy.html - Read 'The Recorded Sound Sucks - We're trying to make it better'. Written by Walter Sear. All of his articles are great to read honestly for someone with an interest in our craft.
This is only opinion with zero science to support it. That interview actually very clearly shows he did not understand sampling theory. He says "The bandpass, the frequency response - much, much too limited. It will improve as they go up to 192. The follow the dots becomes less follow the dots. There are more dots." and "and until the dots get close enough so that the ear can't hear the difference. It's like the old dot matrix printers - you could tell it was an "O" with the little steps up and the little steps down - you could read it as an "O" - but it wasn't very elegant." That was utterly clueless. The guy had no idea how sampling works.
I have colleagues, who went to highly esteemed audio engineering schools, who's instructors could recognize what sample rate projects were recorded in a heart beat.
More anecdotal evidence. It is meaningless. I have heard many many people claim they can hear differences. Once put to the test, they all fail. (And not just with me. With any proper test ever performed anywhere).
Just because you can't hear a difference between a 44.1khz/192khz recording (and your outboard gear's ability to reproduce this response) and their impact on the recorded product, doesn't mean there isn't a perceivable difference to those of us that can.
Ah yes that childish argument. Question my hearing. I could also stoop to that level and call you delusional but what is the point? All science and all evidence points to higher sampling rates being indistinguishable from 44.1Khz. On the other hand expectation bias and placebo are something every single person on the planet is subject to. There is no question about this. 99% of science is about procedures to avoid and work around the flawed human bias. Anyone that claims different needs to come with some very serious and solid science based evidence. Extraordinary claims need extraordinary proof.

On my quality converters, (Prism Sound), no one can tell the difference between different sample rates in a double blind ABX test. That includes many people that told me they could easily tell the difference. Some claimed night and day differences or other such superlatives... that is usually a clear give away that either their converters are completely broken or they have an over active imagination.

I'll side with the AES and established (highly regarded) professionals in our field on this one.
I'll side with A) the scientists and B) the actual proof and science. But you know what what? Most papers at the AES agree with the science and the proof.  (Note that having done a presentation at the AES does not mean that the AES agrees with your conclusions). I don't care how many grammys someone has, their claims are meaningless unless they have solid science and real testing to back-up them up. It is relatively easy to setup a proper double-blind test to demonstrate this yet so far no one, absolutely no one, has managed to. I think the conclusions are more than obvious.

UnderTow
why do you always drag me into this undertow..lmoa..   :)
i was speaking theoreticly...(see i dont eve know how to spell it...theoreticly?)..anyway i said three words..lol..but i do like your arguement..(you know  i always do)
so if this guy was packing amo..then your packing a convey of troops with this post..nice work.

Windows 8.1 X64 Sonar Platinum x64
Custom built: Asrock z97 1150 - Intel I7 4790k - 16GB corsair DDR3 1600 - PNY SSD 220GB
Focusrite Saffire 18I8 - Mackie Control
   
#42
BEATZM1D10T
Max Output Level: -85 dBFS
  • Total Posts : 285
  • Joined: 2009/05/22 12:43:50
  • Location: Mid-West
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 13:45:05 (permalink)
UnderTow


snip

You don't disappoint.

http://www.promastering.com/pages/techtalk_mac/tt-3_mac.html#All%20About%20Sampling

"Rupert Neve does a test where he changes sine waves to square waves with high fundamentals, and people can hear the difference when they should not theoretically be able to, as the only difference is in harmonics that are above the commonly accepted audible range. He also tells a story of Geoff Emmerick correctly pointing out a couple of improperly terminated channels just by listening to the console output when the differences were only a few db down at around 50 kHz. In both cases above, there may be other distortions at work that explain the differences heard, but it remains interesting nonetheless. "

"Even if you discount the contested evidence on human perception of ultrasonic frequencies, to ensure coverage of the entire population, you still need to cover a 24 kHz bandwidth according to the studies, plus leave room for gentler filter slopes, and a bit of space to ensure that the filters won’t have audible artifacts due to ripple. At the very least, you still need 60 - 64 kHz sample rates according to most studies and industry task groups. Interestingly, the committee on sample rate in the 70’s had suggested a 60 kHz sample rate, but for practical reasons having to do with the available technology at the time, the 44.1 and 48 kHz rates were settled upon."

All by Jay Frigoletto . Oh wait...he's one of those 'Grammy Winners' that you don't pay any attention to. I guess thousands upon thousands of hours behind the mixing board and being respected by your peers amounts to nothing...
 
My point bringing up the Nyquist plot is that the HF content gets wrapped down into the audible band unless it's filtered out, and then you miss some of the musical content.

If higher sampling rates are useless why do your precious Prism Convertors offer the option of operating at these useless rates? Just for marketing kicks?

I'll trust the ears of professionals who demonstrate incredible work, time and time again. That's just going to be our difference of opinion here.

Most of the public can't hear the difference between a Marshall 1986 or Marshall 1987. It doesn't matter how many people listen to your A/D's if they don't know what to listen for.

mike_mccue


Highly esteemed audio engineering schools?

Yes Mike. Highly esteemed.
#43
WDI
Max Output Level: -54.5 dBFS
  • Total Posts : 2069
  • Joined: 2007/08/28 02:31:11
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 14:12:28 (permalink)
UnderTow


WDI

If the 192 MP3 still sounds like garbage there most likely is another problem.
Oh absolutely. I am just a stickler for technical details. :-)





UnderTow

Haha, funny cartoon. We all can probably replace the stick figure with a picture of ourselves at times. I'll leave the technical details up to you guys who can argue the details much better then someone like me who gets overwhelmed. The one thing I've found with digital, I have a much harder time hearing differences than with analog equipment. For instance in the digital domain bit depth/sample rates, Edirol FA-66 vs Fireface or plugins such as compressors etc. Seems like in the analog world, each piece of equipment has its own character easy to hear. In the digital domain, primarily with DAWS, I'm usually looking for no crashes and good performance CPU wise and well thought out tools. LOL.

SteveXN sounds good to me! One thing I would note though, I haven't put anything on SoundClick in a long time so I could be wrong, but your 192 MP3 file you sent them may have been converted to a different format of less quality by SoundClick. So if anything, the file you sent them should sound even better. So it seems to me you did everything correct. As far a Boost11, I'd just make sure not to go overboard with the settings. It can be good for getting the overall mix levels were you want them, but it can also easily sound over compressed and harsh. But your mix doesn't sound over compressed to me.

Where did you get a $75 upright bass? Need one of those! LOL!
post edited by WDI - 2011/04/15 14:39:33

Sonar 7 PE
Windows XP Pofessional (SP3)
MSI K8N Neo4-F
AMD Athlon 64 3500+
2 GB PC 3200 Ram
RME Fireface 800
Edirol FA-66
CM Labs MotorMix

Old stuff: ARJO
#44
John
Forum Host
  • Total Posts : 30467
  • Joined: 2003/11/06 11:53:17
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 14:43:49 (permalink)
My point bringing up the Nyquist plot is that the HF content gets wrapped down into the audible band unless it's filtered out, and then you miss some of the musical content.
This has been asserted before with no evidence. I can understand how on a string instrument that a higher string can cause sympathetic vibrations in a lower string but that is due to some sort of coupling. Either with air or a mechanical connection. There is no such coupling with digital audio. To think that a higher frequency has any effect on a lower one in the digital realm is absurd. We are talking about harmonics after all.

I am in complete agreement with what Undertow has said. Further this has been discussed in the old forum a great deal. No one has stepped forward to prove this assertion. 

you still need to cover a 24 kHz bandwidth according to the studies
What studies?

BTW as we age we loose high frequency acuity. We loose the ability to hear those frequencies above 10 to 15 kHz. 20 kHz is accepted as the cutoff for human hearing in the best case i.e young ears. This does not take into account of an audio  engineer working for years in a high volume listening environment and the loss of their high frequency hearing. The last people anyone should use to determined hearing ability are audio engineers. Much as rock musicians have severe hearing loss. I would never ask any of them to listen to say monitors for their opinions.    

If higher sampling rates are useless why do your precious Prism Convertors offer the option of operating at these useless rates? Just for marketing kicks?
Yep it is marketing. If one hardware manufacture puts out one than they all have too.




People can quote anyone they wish as anecdotal evidence which is not proof. 

Best
John
#45
LANEY
Max Output Level: -64 dBFS
  • Total Posts : 1350
  • Joined: 2010/12/11 20:27:13
  • Location: USA
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 14:48:00 (permalink)
;)

you guys are too funny!



i7/16GB ram
Win 7 x64
SONAR Platinum Producer x64
VS-700 C&R

Octa-Capture and VS-100 for live recording
#46
UnderTow
Max Output Level: -37 dBFS
  • Total Posts : 3848
  • Joined: 2004/01/06 12:13:49
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 15:16:59 (permalink)
BEATZM1D10T


UnderTow


snip

You don't disappoint.
Yet you don't actually acknowledge any of the points I made....
http://www.promastering.com/pages/techtalk_mac/tt-3_mac.html#All%20About%20Sampling
This article is outdated. Jay wrote it in 1998. Let's look at a few things the article says but first let me point out that there is no such thing as an analogue linear phase filter so Jay got at least one thing wrong. Anyway:

         "Some expensive digital EQs currently offer an option called “linear phase” which delays all frequencies by the same amount               (There also exist analog linear or minimum phase designs). This linear phase distortion isn’t too bad usually. It just introduces          a short delay into the signal. The larger problem is what we find in more than 99% of all the other filters in the world

This is outdated information. While it might still be true that 99% of filters in general are not linear phase, in modern converters I would wager that 99% of the digital anti-aliasing filters are capable of linear phase response. Often this is user selectable. (Meaning the people that buy the chips and use them in their ADC/DAC designs. I don't mean Joe end-user). As for expensive, these days, the very best converter chips, as a whole (so not just the filter) cost a few dollars a piece. Literally.

The article goes on mentioning other issues with analogue filters. All completely irrelevant considering the frequencies they are actually working at. See below.
        
             "So even if you are of the opinion that the information in music above 20 kHz is completely unimportant, and I’m not sure I               totally agree with this, it becomes obvious that a 44.1 kHz sample rate is inadequate for transparent audio reproduction                    based solely on the real world problems associated with the anti-aliasing filter. It should be mentioned that one way that               current state of the art addresses filter problems is with oversampling. If one initially samples at a rate that is several                         multiples higher than the target rate for the audio, one can relax the filter requirements in the initial sampling, and convert               this high rate audio to the desired sample rate using linear phase, equiripple digital filters. It's not a perfect solution, but                    oversampling and current converter design is a subject unto itself, so we'll leave it there for the moment."

Actually the truth of the matter is that all modern converters (with the exception of a few crappy "audiophile" designs) are sigma-delta modulator designs with the sampler working at 64x or 128x the base rate. That is 5.6 Mhz! It is very easy to design an analogue anti-aliasing filter that let's everything through at 20Khz but nothing at 2.8 Mhz. (Half the sampling rate a.k.a the Nyquist frequency). The rest of the filtering is done in the digital domain and invalidates all the arguments against 44.1 Khz sampling that have so far been presented in the article.
"Rupert Neve does a test where he changes sine waves to square waves with high fundamentals, and people can hear the difference when they should not theoretically be able to, as the only difference is in harmonics that are above the commonly accepted audible range.
Rupert Neve is an absolutely great engineer of analogue equipment but he is not really versed in digital audio. Just dropping the Neve name doesn't really mean much. This is still anecdotal evidence with zero description of how the experiment was done.
He also tells a story of Geoff Emmerick correctly pointing out a couple of improperly terminated channels just by listening to the console output when the differences were only a few db down at around 50 kHz. In both cases above, there may be other distortions at work that explain the differences heard, but it remains interesting nonetheless. "
Yet more anecdotal evidence. Even assuming Geoff Emmerick did hear a difference, which we really can not be sure of, it could be explained by the phase shift that would occur in the audible range in the situation described above. The above is a bit like concluding that someone can see in the ultra-violet range because a slight shift was measured in the ultra-violet range but completely ignoring that in the visible range the object being observed went from dark green to bright yellow. ;-)
"Even if you discount the contested evidence on human perception of ultrasonic frequencies, to ensure coverage of the entire population, you still need to cover a 24 kHz bandwidth according to the studies, plus leave room for gentler filter slopes, and a bit of space to ensure that the filters won’t have audible artifacts due to ripple. At the very least, you still need 60 - 64 kHz sample rates according to most studies and industry task groups. Interestingly, the committee on sample rate in the 70’s had suggested a 60 kHz sample rate, but for practical reasons having to do with the available technology at the time, the 44.1 and 48 kHz rates were settled upon."

This is all based on very old converter designs with analogue anti-aliasing filters working at the base rate. This is all meaningless when considering any converters made in the last decade (or maybe even longer). Although I would not object to using a 60Khz sample rate and standardizing to that.
All by Jay Frigoletto . Oh wait...he's one of those 'Grammy Winners' that you don't pay any attention to. I guess thousands upon thousands of hours behind the mixing board and being respected by your peers amounts to nothing...
Jay is a great audio engineer and does a good job of moderating the mastering forum at Gearlsutz but that does not make him an expert on converters. But yeah, are you sure he still thinks the same as he did in 1998 when he wrote that article? ;-)
My point bringing up the Nyquist plot is that the HF content gets wrapped down into the audible band unless it's filtered out, and then you miss some of the musical content.
Ho! Easy on the logical leaps! What are you basing this "and then" on? Why do you miss musical content if inaudible frequencies are filtered out?
If higher sampling rates are useless why do your precious Prism Convertors offer the option of operating at these useless rates? Just for marketing kicks?
Yes. And compatibility. But there is another reason: In doesn't cost much more to offer higher sampling rates so why not do it? It seems to impress some people. :-)
I'll trust the ears of professionals who demonstrate incredible work, time and time again. That's just going to be our difference of opinion here.
But what are you trusting? What exactly are you basing your point of view on? Outdated articles? The fact that some will offer higher sampling rates, possibly just to please their clients? And when did they participate in proper testing to check if they really could hear a difference? Or are you imbuing these people with mythical super-human powers of objectivity? These people are regular humans you know. Go over to Gearsltuz, you can have a chat with Jay if you want. Maybe you can check what his point of view on sampling rates is 13 years after that article you linked...

EDIT: PS: If I wanted to learn more about the finer details of airplane aerodynamics, I wouldn't ask a pilot. I would ask a scientist. IMO you are looking to the wrong "expert" for your information.
Most of the public can't hear the difference between a Marshall 1986 or Marshall 1987. It doesn't matter how many people listen to your A/D's if they don't know what to listen for.
I am speaking of colleague sound engineers but more importantly, no one anywhere in the world with any level of experience, any number of grammys etc etc has ever been able to irrefutably demonstrate in a double blind test that they can hear the difference of higher sampling rates. Zero, zip, nada! So who are you agreeing with exactly? And what are you agreeing with?

But it's cool, I'll agree to disagree. Just please don't pretend there is any scientific basis for higher sampling rates for recording. (Processing is another issue).

UnderTow
post edited by UnderTow - 2011/04/15 15:25:25
#47
The Maillard Reaction
Max Output Level: 0 dBFS
  • Total Posts : 31918
  • Joined: 2004/07/09 20:02:20
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 15:48:55 (permalink)

mike_mccue


Highly esteemed audio engineering schools?

Yes Mike. Highly esteemed.




Will we ever learn the name of these institutions you esteem so highly?







#48
The Maillard Reaction
Max Output Level: 0 dBFS
  • Total Posts : 31918
  • Joined: 2004/07/09 20:02:20
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 16:14:33 (permalink)
"... All by Jay Frigoletto . Oh wait...he's one of those 'Grammy Winners' that you don't pay any attention to. I guess thousands upon thousands of hours behind the mixing board and being respected by your peers amounts to nothing... "

Jay hasn't won a Grammy and Jay never said he won a Grammy either.

Roger a.k.a. BeatzM1d1ot is telling me that Jay has won a Grammy... but now I'm questioning Roger's credibility.

Last week Roger suggested that I might be deaf... he wasn't really sure... but he thought suggesting it as a possibility worthy of mention. I responded by explaining that I have thousands of hours of mixing experience and 30 years of working audio tech and that I still get hired by my peers... but that didn't seem to mean anything to him.

:-)


Here's the Grammy winning album Jay worked on in 2004:

2004
Best Mexican/Mexican-American Album: Intimamente

artist: Intocable (José Juan Hernández, René Martinez, Ricardo Muñóz, Félix Salinas, Daniel Sanchez, Sergio Serna)

engineers/mixers:  Jack Saenz & Malcolm Harper, Jr.





post edited by mike_mccue - 2011/04/15 16:16:17


#49
tarsier
Max Output Level: -45 dBFS
  • Total Posts : 3029
  • Joined: 2003/11/07 11:51:35
  • Location: 6 feet under
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 16:16:14 (permalink)
It's interesting that BEATZM1D10T would bring up Rupert Neve and Walter Sear in a discussion about digital audio. I attended a lecture by Neve where he repeated that Geoff Emerick story. It seems to be one of his favorites, and he tells it to illustrate why he designs his analog products to have a very wide frequency response.  But he also admitted that he knew very little about digital audio--he's a total analog guy.

I also attended an open house that Walter Sear gave at his studio during one of the AES conventions. It was fascinating, especially to hear all the misinformation he said about digital audio. He's another fantastic analog guy, but it was quickly obvious that he didn't have a good grasp of digital audio.
#50
bitflipper
01100010 01101001 01110100 01100110 01101100 01101
  • Total Posts : 26036
  • Joined: 2006/09/17 11:23:23
  • Location: Everett, WA USA
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/15 18:46:17 (permalink)
I'd jump in here, but the last 27 times we had this discussion wore me out.

Geez, is it so hard to just read a book or two before pronouncing yourself an expert?


All else is in doubt, so this is the truth I cling to. 

My Stuff
#51
portesham
Max Output Level: -88 dBFS
  • Total Posts : 129
  • Joined: 2005/01/04 08:32:16
  • Location: Weymouth, Dorset, UK.
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/16 07:27:00 (permalink)
SteveXN


If anyone wants to hear the 192 bit version it is here. I was going to do some vocal retakes, but it may be interesting to you as is

http://www.soundclick.com/player/single_player.cfm?songid=10524472&q=hi&newref=1


Nice! Did you record in one take, and if so, how did you mic up your vocal and guitar? Good open sound!
 
Also, as I don't like MP3, consider/experiment with other file formats even if the ultimate target is internet or CD.

Sonar Professional on Win 10, Core 2 quad Q66000, 4GB ram, Tascam US-2x2, NVidia 9400GT, A-500Pro.

Excelling at mediocrity for 61 years.
#52
SteveXN
Max Output Level: -89 dBFS
  • Total Posts : 63
  • Joined: 2007/12/01 12:11:22
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/16 09:33:11 (permalink)
I like your tag "excelling at mediocrity..." I like to say that in a world of black and white, I'm searching for the grayest gray.
 
I'm pretty ignorant. I sat with the ksm27 large dia about 10" in front of my face (12" above sound hole) and a newly purchased sm81 about 2" off the 12th fret facing towards the sound hole, lead voc and guitar in one shot, both mics catching both sources. Eq is just a 5db bump at 100hz on the guitar, with same on vocs plus some lift at 5k and 15k. I put a touch of phase shift on guitar 12deg/113width on this and use a bus for lexicon pantheon set to stock "room" preset, level about -12 on both sources. 10db compression on both voc and guitar. Overdubbed bass and harmony. My poor bass is 18db compression with a 10db eq boost at 60hz and everything over 1khz rolled off 10db

These are the works of man? This is the sum of our ambition? Sting

Sonar Platinum, Sapphire Pro40, Gateway DX4840, i3 550 @ 3.20GHz, 64 bit, 8GBRam, Shure KSM27/SM81, Takamine
#53
aaronbewza
Max Output Level: -89 dBFS
  • Total Posts : 65
  • Joined: 2009/10/02 04:43:25
  • Location: Cache Creek, British Columbia, Canada
  • Status: offline
Re:MP3 encoder sounds terrible 2011/04/18 05:20:52 (permalink)
Or (notwithstanding the excellent advice by many valued members), you could try using my patch to add all possible Mp3 entries directly to Sonar's Export Audio menu and then pick-and-choose from 44 available bitrate options, until you find a bitrate setting that suits your needs. You can also choose to keep (or throw away) your wav file. There are, well, all of them to choose from... constant and variable bitrates. I'd suggest using variable bitrate at its best setting, or at least 128kbps at constant bitrate.
 Information is in this thread:
http://forum.cakewalk.com/tm.aspx?m=2268643

Hope this helps your problem with icky-sounding Mp3s... cheers!
btw there's no messing with the external encoder utility this way, at all.
Aaron
post edited by aaronbewza - 2011/04/18 05:24:14

You never get a third chance to make a second impression.
#54
Page: < 12 Showing page 2 of 2
Jump to:
© 2024 APG vNext Commercial Version 5.1