Post by: cerberus on October 14, 2007, 11:40:05 PM
please help me with facts and opinions on the issue of mp3s that
are converted to pcm... and they are all clipping!
does it really happen?
t.i.a.
jeff dinces
Post by: jdg on October 15, 2007, 12:53:49 AM
i know its fact that converting to mp3 from a pcm source close to 0dBFS will clip when transcoded (played) back
Post by: Tomas Danko on October 15, 2007, 04:12:37 AM
| cerberus wrote on Mon, 15 October 2007 04:40 |
http://recforums.prosoundweb.com/index.php/m/281284/2119/#ms g_281178 please help me with facts and opinions on the issue of mp3s that are converted to pcm... and they are all clipping! does it really happen? t.i.a. jeff dinces |
Yes, the way the audio is encoded dynamically it's bound to clip upon decoding.
It creates it's own approximate scale trying to hit the original dynamical targets. Obviously it will sometimes go below, sometimes above. Trying to iterate until it's almost spot on takes a lot more time when encoding, and in reality it's not going to hit the spot anyway.
Which is why we should top out at -0.3 dBfs to avoid the worst and obvious clipping of mp3's.
Post by: cerberus on October 15, 2007, 08:02:08 AM
| Tomas Danko wrote on Mon, 15 October 2007 04:12 |
... in reality it's not going to hit the spot anyway. |
i feel bad knowing i have sent mp3s to clients that clipped. though i suppose that
most of the stuff for sale at the itunes store clips too.
like, where is craig anderton when we need him?
imo, we've got to make people aware, or this situation won't improve.
jeff dinces
Post by: Fede on October 15, 2007, 11:51:03 AM
Post by: Tomas Danko on October 15, 2007, 03:49:40 PM
| Fede wrote on Mon, 15 October 2007 16:51 |
Not that I buy a ton from the iTunes store (would much rather buy a CD and encode at lossless, as I can rarely stand MP3) and yes I have probably ranged towards singer/songwriter (although I have bought some hip-hop) rather than truly loud music, but to my knowledge nothing I have bought from there has notably clipped. |
Since the files bought from iTunes are protected, how do you open them into an editor to see the real audio and if it's clipping or not, instead of just a flat line?
Just being curious.
Post by: bblackwood on October 15, 2007, 03:51:21 PM
| Tomas Danko wrote on Mon, 15 October 2007 14:49 | ||
Since the files bought from iTunes are protected, how do you open them into an editor to see the real audio and if it's clipping or not, instead of just a flat line? Just being curious. |
Burn CD, import into editor?
Post by: Tomas Danko on October 15, 2007, 04:04:13 PM
| bblackwood wrote on Mon, 15 October 2007 20:51 |
Burn CD, import into editor? |
Sssshhhh, I asked him not you. Now he will know one way to answer the question!
Just wondering if Fede had actually checked those files, or relying solely on listening tests. (Which in the end is what matter, but we're talking mastering geek stuff right now)
Post by: bblackwood on October 15, 2007, 04:06:26 PM
| Tomas Danko wrote on Mon, 15 October 2007 15:04 | ||
Sssshhhh, I asked him not you. Now he will know one way to answer the question! |
Sorry.
*bows head and stands in corner*
Post by: Nigel Jopson on October 16, 2007, 01:48:47 PM
| cerberus wrote on Sun, 14 October 2007 22:40 |
mp3s that are converted to pcm... and they are all clipping! jeff dinces |
... well, maybe it has to do with distortion introduced when converting from PCM to MP3 in the first place!
At last year's AES Paris conference I chaired a seminar in which we isolated distortion components from MP3s. First we played the audience the MP3, then the distortion. Some people took their IR headphones off when the distorted extracts were played. The tunes were encoded from a "name and shame" list of hot CD masters Thomas Lund (of TC Electronic) had compiled in an article (Distortion To The People) he wrote for Resolution magazine:
http://resolutionmag.com/pdfs/TECHNO~1/DISTOR~1.PDF
Some work which the perceptual codec distortion investigations were based on is included in this paper:
http://www.tcelectronic.com/media/nielsen_lund_2003_overload .pdf
Just as interesting, to me, is the distortion generated by loud-but-legal levels in consumer players. Modern upsampling players can generate 3-10% distortion from perfectly "legal" levels of test signals. Thomas had tested a range of players owned by TC Electronic staff. This part of his talk is in this excerpt:
http://www.jop.myzen.co.uk/arc/2006_aes_Lund.mov
When I'm mastering I always check the 'test burn' on a range of ordinary players, and I think there's definitely a good case to be made for data-reducing encoding to be done carefully by the mastering engineer ...
Post by: compasspnt on October 16, 2007, 01:57:28 PM
| Nigel Jopson wrote on Tue, 16 October 2007 13:48 |
Just as interesting, to me, is the distortion generated by loud-but-legal levels in consumer players. |
YES!!!
Thank you.
Post by: Alexey Lukin on October 17, 2007, 02:46:50 AM
| cerberus wrote on Sun, 14 October 2007 23:40 |
please help me with facts and opinions on the issue of mp3s that are converted to pcm... and they are all clipping! |
Yes, they are clipped if levels of the original PCM were close to 0 dBFS. This clipping is not introduced during encoding, it's introduced during decoding. The reason is that MP3 format, just as all other lossy formats, distorts the waveform. And any waveform distortion near 0 dBFS has a chance of going over 0 dBFS. Since most programs decode to fixed-point PCM, the result will be clipped.
If you are using RX on a mac, it will decode MP3 to floating-point PCM, and you can get the non-clipped decoding, suitable for normalization or limiting. This doesn't work on a PC yet! (the Windows WMA decoder that we are currently using for MP3 can only decode to fixed-point).
Post by: Matt_G on October 17, 2007, 07:43:01 AM
Matt
Post by: cerberus on October 17, 2007, 09:39:06 AM
i did indeed use rx (mac) in the manner you've outlined. once i learned the
true reconstructed signal values, it was an eye opening and educational
experience for me. the values were way outside of my expectations.
in fact it was reminiscent of figuring out that santa claus does not exist.
most kids play along for a while after they discover the truth because
it is to their benefit, in the end though, poor kids get nothing.
that was the hardest part to admit. well.. there may be no
santa... but -they- have a "santa", don't they? sadly, no.
jeff dinces
Post by: Ben F on October 17, 2007, 09:10:55 PM
Post by: cerberus on October 18, 2007, 12:20:38 AM
| Ben F wrote on Wed, 17 October 2007 21:10 |
Just master to -0.3dBFS and you should be covered for most codecs. |
that MYTH is so nice and convenient. thank you for proving my point.
you and everyone are cordially invited to discover the truth.
jeff dinces
Post by: Tomas Danko on October 18, 2007, 06:07:25 PM
| cerberus wrote on Thu, 18 October 2007 05:20 | ||
that MYTH is so nice and convenient. thank you for proving my point. you and everyone are cordially invited to discover the truth. jeff dinces |
Myth or not.
It's just that for people who do not wish to investigate the matter further this will improve things quite a lot compared to the worse alternatives without the engineer having to spare any more thoughts about it all.
It's good advice, the way I look at it. Heck, it's even backed up by certain tests and numbers from Mr Paul Frindle. I mean, how bad can it be...
Post by: Ben F on October 18, 2007, 09:28:07 PM
| cerberus wrote on Thu, 18 October 2007 13:50 | ||
that MYTH is so nice and convenient. thank you for proving my point. you and everyone are cordially invited to discover the truth. jeff dinces |
The truth is none of my clients have ever complained, and that's good enough for me. I do far more complex encoding/decoding infrastructure for broadcasting so it's not really on the top of the priority list.
Post by: cerberus on October 19, 2007, 01:30:29 AM
the defensive. it is not any mastering engineer's fault that ca. 2007:
A. mp3 became the predominant delivery format for music.
B. music sales dropped by 15% over the past ten months.
---
i see an opportunity now for the mastering community to
suggest/promote a better delivery format for music.
like... one that the fans would pay for.
jeff dinces
Post by: Bob Olhsson on October 19, 2007, 08:58:00 PM
The other thing about lossy coders is that they all introduce a 1/3 octave filterbank into the equation both going in and coming out. Filters ring when they get clipped.
My wife wanted to put a tune up on myspace so we did a shootout between samplitude and i-tunes. The latest i-tunes seems to drop the level!
Post by: Alexey Lukin on October 22, 2007, 04:13:07 AM
| Matt_G wrote on Wed, 17 October 2007 07:43 |
What sounds worse the clipping on decode to PCM or the Mp3 compression? |
It depends on amount of clipping. For many mp3 files with hot levels, quality can be significantly improved by non-clipped decoding, esp. when bit rates are low.
| Matt_G wrote on Wed, 17 October 2007 07:43 |
If consumers don't care about listening to uncompressed audio why worry whether it's clipping? |
Because in uncompressed audio clipping is controlled by a mastering engineer. In case of mp3, clipping is usually not controlled by anyone caring about sound quality.
Post by: Alexey Lukin on October 22, 2007, 04:16:25 AM
| Ben F wrote on Wed, 17 October 2007 21:10 |
Just master to -0.3dBFS and you should be covered for most codecs. |
Not quite. For medium bit rates, dropping levels by 0.3...0.5 dB really improves the result, but it's not always sufficient: clipping can get higher than that.
Post by: Matt_G on October 22, 2007, 11:13:02 AM
| Alexey Lukin wrote on Mon, 22 October 2007 18:13 |
It depends on amount of clipping. For many mp3 files with hot levels, quality can be significantly improved by non-clipped decoding, esp. when bit rates are low. |
So the solution is to produce a decoder that eliminates the clipping while keeping the levels consistent with the uncompressed PCM version. Then get Apple to license this product for playback in iTunes/iPods. Is iZotope RX capable of doing this?
| Quote: |
Because in uncompressed audio clipping is controlled by a mastering engineer. In case of mp3, clipping is usually not controlled by anyone caring about sound quality. |
My point exactly, if the consumer can't hear lossy Mp3 encoding are they going to hear clipping? If we lowered the level by 6db to prevent the decoder clipping they would certainly notice the level drop.
Matt
Post by: Bob Olhsson on October 22, 2007, 02:22:55 PM
Post by: Tomas Danko on October 23, 2007, 05:48:53 PM
| Bob Olhsson wrote on Mon, 22 October 2007 19:22 |
It's utterly silly to expect an ordinary listener to identify why they don't like the sound of something. If people don't like what they hear, they simply don't listen again, much less buy the CD or file. Digital clipping causes gross degradation from subsequent digital processing such as adjusting the digital volume control that is found on virtually every modern music player. |
Very well said!
Post by: Alexey Lukin on October 23, 2007, 06:17:54 PM
| Matt_G wrote on Mon, 22 October 2007 11:13 |
Then get Apple to license this product for playback in iTunes/iPods. Is iZotope RX capable of doing this? |
Yes, it is, but it's actually using Apple QuickTime to decode mp3. So, Apple iTunes (and all other players) just needs a good mastering limiter to bring peak levels back below 0 dBFS.
Post by: Tomas Danko on October 23, 2007, 06:30:42 PM
| Alexey Lukin wrote on Tue, 23 October 2007 23:17 | ||
Yes, it is, but it's actually using Apple QuickTime to decode mp3. So, Apple iTunes (and all other players) just needs a good mastering limiter to bring peak levels back below 0 dBFS. |
Wouldn't it be too late for the limiter to remedy this problem after the decoder has already clipped the signal? And wouldn't a non-linear process such as a limiter make things even worse in such a case?
Lowering the source material level just a tad bit would make all that a moot point anyway.
Post by: Alexey Lukin on October 24, 2007, 02:59:40 AM
Post by: Tomas Danko on October 24, 2007, 04:16:22 AM
| Alexey Lukin wrote on Wed, 24 October 2007 07:59 |
The decoder can work in a non-clipping mode if players use its 32-bit FP decoding feature. |
Yes, and that would be ideal. But it turns out very few decoders are doing this, in reality.
Post by: cerberus on October 24, 2007, 08:41:12 AM
i suppose it is for the itunes store.
i feel so dirty.
jeff dinces
Post by: Matt_G on October 26, 2007, 12:16:32 AM
| Alexey Lukin wrote on Wed, 24 October 2007 08:17 | ||
Yes, it is, but it's actually using Apple QuickTime to decode mp3. So, Apple iTunes (and all other players) just needs a good mastering limiter to bring peak levels back below 0 dBFS. |
I understand the RX decoder is 32bit float & that it has the necessary headroom to expand the signal on decode beyond 24bits without clipping, but what's the point if you have to strap another limiter on the end of it prior to D/A conversion?
Then we get into the age old debate, what sounds better clipping or limiting. Not to mention all the additional DSP processing from RX & the limiter & do we need to dither... blah, blah. My guess would be that clipping would sound closer to the original master then this work around.
Matt
Post by: cerberus on October 26, 2007, 04:04:40 AM
| Matt_G wrote on Fri, 26 October 2007 00:16 |
I understand the RX decoder is 32bit float & that it has the necessary headroom to expand the signal on decode beyond 24bits without clipping, but what's the point if you have to strap another limiter on the end of it prior to D/A conversion? |
| Alexey Lukin wrote on Wed, 17 October 2007 02:46 |
suitable for normalization or limiting. |
pancakes per song (for now). i hope it isn't audible (example below).
| Quote: |
Then we get into the age old debate, what sounds better clipping or limiting. ... My guess would be that clipping would sound closer to the original master then this work around. |
yield the results that are most similar to the original master.
| cerberus wrote on Wed, 24 October 2007 08:41 |
another request for a loud 192 mp3 last night. i suppose it is for the itunes store. i feel so dirty. |
and that is what i get paid for. here it is. the analog reconstruction will clip at five individual samples; maybe six if the
converter is cheese. could anyone hear that? does it affect the listener? i don't know.
jeff dinces
Post by: Matt_G on October 26, 2007, 07:53:13 AM
| cerberus wrote on Fri, 26 October 2007 18:04 |
matt has just convinced me to allow five discontiguous samples into the house of pancakes per song (for now). i hope it isn't audible (example below). |
Haven't listened but did you hear anything? it's up to you & your client to be the judge, use your ears. If the overs are on quick transients you won't hear it.
| Quote: |
attenuating the gain and then making it up in the analog domain would yield the results that are most similar to the original master. |
Not if you were level matching the attenuated Mp3 with the original pcm master. As I said I doubt a handful of transient peak overs are going to be audible, certainly not as audible as the Mp3 encoding space monkeys
Matt
Post by: cerberus on October 26, 2007, 09:34:53 AM
| Matt_G wrote on Fri, 26 October 2007 07:53 |
Haven't listened but did you hear anything? it's up to you & your client to be the judge, use your ears. If the overs are on quick transients you won't hear it. |
one's food, err... a few grains here and there won't harm you. likewise, i believe
that the presence of white noise in music has an "irritating" or disturbing"
psychoacoustic effect. this may trigger certain adrenaline centers too,
perhaps makes the music exciting. like chrome on a corolla.
hot peppers to make boring food seem exciting.
imo, whether a test subject can identify the source of the nervous irritant is another
question. if they can't, it still doesn't mean that there is no psychoacoustic effect.
hot peppers taste spicy; eat enough and they will make you perspire. so i think
that likewise: asking a music listener what they "hear"does not cover all.
music is like a drug. it can cause the brain to produce endorphins, etc.
(but not violent behavior, that would be tv and videogames! :+)
as for the minimum duration for this effect, i do not know. however, we could say that
the clipped signal is -loud-. why could we assume that would ever be inaudible? and
considering that most playback systems would not recover in time to reproduce
the immediately following samples accurately, the effect on reproduction in the
analog domain could last longer than it may appear to on our screens.
| Quote: |
As I said I doubt a handful of transient peak overs are going to be audible, certainly not as audible as the Mp3 encoding space monkeys |
any kind of torture causes unbearable discomfort. we cannot say one kind
of torture is better than another. and as i said above. i am not so
concerned with audibility as in the effect on brain chemistry or
neuro-electrical activity. on mood, which affects behavior.
even if a sound is masked; it still affects everything.
| Matt_G wrote on Fri, 26 October 2007 00:16 |
Then we get into the age old debate, what sounds better clipping or limiting. |
default now. i think that music fans decided. i think we could plot these
trends on a graph, the total number of samples clipping on major
label releases would correlate with the drop in sales. thus my
theory. just ask anyone: "do old recordings sound better?"
jeff dinces
Post by: Patrik T on October 26, 2007, 03:55:32 PM
I can't believe why there is such a mystery to this!
Balance. Few chopped of tops, or hopefully none. Good room. Good ears.
Best Regards
Patrik
Post by: Sonovo on October 26, 2007, 04:08:35 PM
| cerberus wrote on Fri, 26 October 2007 10:34 |
music is like a drug. it can cause the brain to produce endorphins, etc. (but not violent behavior, that would be tv and videogames! :+) jeff dinces |
I would say that certain forms of music as well as certain rhythms do indeed promote violent behaviour.
A few years ago during Carnival in Brazil, when one particular band played, people went crazy, violence/fights erupted everywhere, etc. As soon as they passed and the next band arrived, the entire atmosphere changed completely. Still full of energy (and very loud), but completely different (fun/happy).
I asked a few friends there about that, and they said it was common knowledge that with that one band (and with slightly different rhythms than anyone else) there was always a lot of fighting and violence. It was just part of the party, you had to take care when they played.
Obviously this could be due to other factors, but f.x. the bands image was one of solidarity and community, they offered many outreach programs, etc. They were by no means known as a 'bad boy' group of players.
There are other musical genres that I think could also be linked to violent behaviour, perhaps one's better known in the industrial world. I'll leave that as an excercise to the reader
Cheers,
Thor
Post by: cerberus on October 26, 2007, 04:23:27 PM
the listener (like a fire alarm), which is the first step toward selling it. but if
the first step prevents the second step...
jeff dinces
Post by: Alexey Lukin on October 26, 2007, 05:58:31 PM
| Matt_G wrote on Fri, 26 October 2007 00:16 |
I understand the RX decoder is 32bit float & that it has the necessary headroom to expand the signal on decode beyond 24bits without clipping, but what's the point if you have to strap another limiter on the end of it prior to D/A conversion? Then we get into the age old debate, what sounds better clipping or limiting. |
Right, it's a debatable question. But in most cases a good limiter will sound better than clipping. This is particularly true on tonal (not percussive) material.
Post by: cerberus on October 26, 2007, 06:49:44 PM
| Alexey Lukin wrote on Fri, 26 October 2007 17:58 |
in most cases a good limiter will sound better than clipping. |
any limiter near the -end- of this chain; so i cannot logically support or
agree with that solution for the playback chain.
since i found that it is possible to cut loud masters which don't clip appreciably when encoded
to mp3; that solution seems to me to be the one that we all ought to pursue.
jeff dinces
Post by: Matt_G on October 26, 2007, 09:02:43 PM
| Alexey Lukin wrote on Sat, 27 October 2007 07:58 | ||
Right, it's a debatable question. But in most cases a good limiter will sound better than clipping. This is particularly true on tonal (not percussive) material. |
I totally agree that a limiter sounds best on tonal or sustained material. Percussive transients is another matter. Ozone covers most of this stuff really well.
Matt
Post by: Alexey Lukin on October 27, 2007, 07:12:23 AM
Post by: cerberus on November 02, 2007, 09:59:21 PM
file. (in alphabetical order according to the host-app name). blue lines represent literal
digital sample values. red lines represent an interpolated analog waveform.
jeff dinces
Post by: cerberus on November 02, 2007, 10:01:54 PM
Post by: cerberus on November 02, 2007, 10:03:42 PM
Post by: cerberus on November 02, 2007, 10:05:16 PM
Post by: cerberus on November 02, 2007, 10:06:42 PM
Post by: cerberus on November 02, 2007, 10:08:12 PM
Post by: cerberus on November 02, 2007, 10:09:26 PM
Post by: cerberus on November 02, 2007, 10:11:05 PM
Post by: chrisj on November 02, 2007, 11:57:45 PM
Lately I've just been refusing to master anything obnoxiously loud.
Good thing I pay my mortgage by writing Audio Unit plugins that are expected to sound GOOD
Nothing _I've_ mastered in the last half a year would cause problems with clipping mp3s. ...or would help pay much of my mortgage...
Enjoy. I'm just glad to be out of it. Such a relief to refuse to care any longer about meeting or beating the horribleness...
Post by: cerberus on November 05, 2007, 08:48:16 PM
jeff dinces
Post by: chrisj on November 06, 2007, 06:06:37 PM
Post by: Ben F on November 07, 2007, 07:37:58 PM
I generally export from Sequoia at 192kbps for clients. As long as there isn't intersample clips and the level is -0.3dBFS I haven't had a problem with clipping.
Post by: cerberus on November 08, 2007, 03:28:41 AM
peak amplitude values. i wouldn't recommend using such limited data as a primary
criterion for determining suitability. in this case, i only made a measurement for
the sake of [going back to make] a gain adjustment before src and dithering.
jeff dinces
Post by: djwaudio on November 08, 2007, 01:11:37 PM
I don't know if it would affect my EQ decisions or not, but I always listen through the final dithering of choice? Why not be able to turn on the MP3 compression to hear what's going to happen when it hits the meat grinder?
Post by: Kees de Visser on November 12, 2007, 06:11:40 AM
| djwaudio wrote on Thu, 08 November 2007 19:11 |
I don't know if it would affect my EQ decisions or not, but I always listen through the final dithering of choice? Why not be able to turn on the MP3 compression to hear what's going to happen when it hits the meat grinder? |
If you're concerned about (low bitrate) lossy codec compatiblity of your audio, the following white paper might be of interest.
I found it quite informative.
http://www.neuralaudio.net/downloads/Codec_Compatibility.pdf
Post by: Roland Storch on November 15, 2007, 04:32:40 PM
| Alexey Lukin wrote on Wed, 17 October 2007 07:46 |
If you are using RX on a mac, it will decode MP3 to floating-point PCM, and you can get the non-clipped decoding, suitable for normalization or limiting. This doesn't work on a PC yet! (the Windows WMA decoder that we are currently using for MP3 can only decode to fixed-point). |
What about Foobar for Windows? I was told it would decode in 32 floating point.
http://www.foobar2000.org/
Post by: cerberus on November 17, 2007, 02:11:57 AM
| Kees de Visser wrote on Mon, 12 November 2007 06:11 |
[ http://www.neuralaudio.net/downloads/Codec_Compatibility.pdf |
away with a big fat -stereo- bass or kick drum. i have used mp3 to try and prove
to clients that such things do not work. the problem may arise that a mix
engineer prefers to claim mp3 sucks; and so does not recognize that
their mix is deeply flawed.
jeff dinces
Post by: Patrik T on November 19, 2007, 10:55:36 PM
The OS is Vista Home Premium and I don't know if digital audio in Windows has become better, but the reults were excellent at 192 kbps.
I might add that the CD is peaking around -6 dBFS (classic chamber music). And I must add that I am no fan of mp3 at all.
But I felt the reults were extremely "good". Ripping this way. I felt absolutely no urge what so ever to rip or encode with something else. Windows Media Player. WTF? Well...
This kinda reminded me that getting softer will save people a lot of time trying to find all kinds of codecs and solutions for everything. Time that can be spent on better things in life.
Best Regards
Patrik
Post by: Tomas Danko on November 20, 2007, 03:51:56 AM
Post by: Patrik T on November 20, 2007, 08:37:36 AM
- 3 dB instead of -0.3. Or no limiter instead of a limiter. Or tiny processing instead of endless processing chains.
The "style" of music is secondary - it's how it GETS there. A bad sounding mp3 begins to crap out at the recording stage and no that much at the end (mastering, whatever) stage.
I simply can not justify myself to spend my precious lifetime on finding solutions for making bad sounding music sound good at low digital resolutions. But many others do spend their days that way, hunting codecs, "solutions" and claiming this and that.
Either it sounds acceptable or not. It is mp3!!! I don't spend time trying to enjoy mp3's through my Lavry converter feeding my fairly flat speakers. I play mp3 through consumer things and how it sounds there is all that should matter.
Best Regards
Patrik
Post by: Tomas Danko on November 20, 2007, 09:36:16 AM
| Patrik T wrote on Tue, 20 November 2007 13:37 |
Sure, sure Mr Danko. But if that rock peaked lower - for starters - then that rock would surely make a better mp3. |
I agree, it surely would.
| Patrik T wrote on Tue, 20 November 2007 13:37 |
- 3 dB instead of -0.3. Or no limiter instead of a limiter. Or tiny processing instead of endless processing chains. |
Again, it would sound even better indeed.
| Patrik T wrote on Tue, 20 November 2007 13:37 |
I simply can not justify myself to spend my precious lifetime on finding solutions for making bad sounding music sound good at low digital resolutions. But many others do spend their days that way, hunting codecs, "solutions" and claiming this and that. |
Once more I agree completely with your point of view. I also just pick a tool and move on with other things.
| Patrik T wrote on Tue, 20 November 2007 13:37 |
Either it sounds acceptable or not. It is mp3!!! I don't spend time trying to enjoy mp3's through my Lavry converter feeding my fairly flat speakers. I play mp3 through consumer things and how it sounds there is all that should matter. |
I do the very same thing regarding mp3 files and consumer things.
With that said, looking at what happens inside the lossy codec will show that a classical piece will suffer a whole lot less than a busy heavy metal track. Limiter or no limiter. Regardless of post processing and mangling at the mastering stage, mp3's do lend themselves better to some music than others, by default.
Regards,
Danko
Post by: Patrik T on November 20, 2007, 10:53:23 AM
Will there be new software companies trying to battle the clipped dragons?
"Smoothness guaranteed!"
"Look here how good we sound!"
Will there be money made?
I guess so.
Cause it all makes the world go around, doesn't it? The fear of sounding "bad".
BR
Patrik
Post by: zmix on November 20, 2007, 01:10:59 PM
Can you run a separate test to determine if these artifacts (below) are the result of the "Smart Encoding Adjustments" or the "Filter below 10hz" processing?
| cerberus wrote on Fri, 02 November 2007 22:03 |
Post by: cerberus on November 20, 2007, 05:58:24 PM
both those options cause some of the distortions. the "smart" causes a greater amount
of "waveform excursions" near the top of the scale. the distortions appear to become
much more severe if the two options are applied together.
jeff
Post by: cerberus on November 20, 2007, 11:52:03 PM
| Patrik T wrote on Mon, 19 November 2007 22:55 |
I might add that the CD is peaking around -6 dBFS (classic chamber music). |
or do you think that your reconstructed analog waveform likely rises nearer to zero?
| Patrik T wrote on Tue, 20 November 2007 08:37 |
I simply can not justify myself to spend my precious lifetime on finding solutions for making bad sounding music sound good at low digital resolutions. But many others do spend their days that way, hunting codecs, "solutions" and claiming this and that... |
and yet despite all that, which seems well and good; i would say that this thread
seems to have truly captured your imagination.
| Patrik T wrote on Tue, 20 November 2007 10:53 |
So where are the logic and natural conclusions? This topic has extended over four pages already. With attached graphs (whassup with that dB-scale btw?). Will there be four pages more? Will there be new software companies trying to battle the clipped dragons? |
i didn't understand your "db scale" question.
jeff dinces
Post by: Patrik T on November 26, 2007, 08:59:10 PM
| cerberus wrote on Wed, 21 November 2007 05:52 |
i didn't understand your "db scale" question. |
Aight.
Well, to my eye it seems your dB scale on the right side of those graphs stretches between...+0.2 and -0.2? Is that around some kind of "ceiling"? Under/over the maximax?
From what I recall, when trying out things, a wav that was limited to -0.6 dB fisted out peaks at around +2 dB when re-imported into a DAW (after encoded to mp3).
On the meters that was. Ear judged things in other ways.
So...logic tells me I'd need something peaking -3 dB (not - 0.3 dB) to ensure there would be no overs in any possible case. Hence the dB-scale question. I mean - why bother measuring things if the scale is not enough to tell what is happening?
Best Regards
Patrik
Post by: cerberus on November 27, 2007, 02:05:10 AM
represents fixed point zero; a.k.a. convertus maximus.
those views are zoomed in on the top section of the rendered waveform to show more
information. such as: that the various mp3 encoders, even the fraunhofer ones,
are -not- identical to each other. this does not affect that they all are
clipping it like a lawnmower.
jeff dinces
Post by: Patrik T on November 27, 2007, 02:32:11 AM
| cerberus wrote on Tue, 27 November 2007 08:05 |
the images depict a zoomed-in view of the top of the waveform. zero on that scale represents fixed point zero; a.k.a. convertus maximus. |
Well, that's what I thought.
Without trying to debate things, just try -3 dB instead of -0.3 dB on the limiter, print a wav of it and make it mp3 and then reimport the mp3 and watch the peak meters in your floating point DAW.
It is surprising how much lossy encoding might rumble a DA.
/Patrik