MP3's..... huge library....Ability to obtain HiFi? - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 12Likes
Reply
 
Thread Tools
post #1 of 37 Old 09-11-2015, 09:29 AM - Thread Starter
Member
 
jlm86's Avatar
 
Join Date: Feb 2003
Posts: 183
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 26 Post(s)
Liked: 10
MP3's..... huge library....Ability to obtain HiFi?

Ok, i'm back after few years away..
Trying to revamp some HiFi sound and see if my huge 320kb ripped mp3 library should just be for the kids..

In General:
Sounds ok listening on computer with Logitech speakers and sub cranked...
The sonics as you go up in HiFi component quality is actually detrimental to listening to these files....

Once i stream this to my HT setup (Onkyo AVR, HSU 5.1 package, PS3) the sound as most of you know is not good.
Also have a HTPC Kodi/Linux setup.. Plex streaming etc., PS3 Player

So, basic questions and thoughts:
1. Is it possible to perform some sonic equalization on this stream to output to a acceptable sound quality?
2. I have some FLAC files, but even on streaming that thin audio mp3 sound is still present.
3. I dumped all my vinyl 10 years ago, tired of moving these albums (my daughter now pissed at me)
4. My hearing is not as good as it was, as most of us can/might attest to...

Thoughts...?

Best..
//
jlm86 is offline  
Sponsored Links
Advertisement
 
post #2 of 37 Old 09-11-2015, 11:57 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
1) 320kbit/sec MP3 files that are encoded by a decent encoder are going to be completely indistinguishable from the CD original to nearly all listeners even on a decent hifi system. However, if you have a dedicated 2 channel listening room with very finely tuned acoustics that can achieve real 3D depth to music the MP3s will not image like the original CD.

2) FLAC files are 100% identical to the CD if ripped correctly. They will sound identical.

There's no magic necessary to send audio files around your house. I'd check that your Plex server isn't recompressing the audio to some low quality / low bitrate MP3 setting, but sending out the original unaltered audio.
Stereodude is offline  
post #3 of 37 Old 09-11-2015, 01:35 PM
AVS Forum Special Member
 
fbov's Avatar
 
Join Date: Nov 2006
Location: Bushnell's Basin, NY
Posts: 1,632
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 196
I'd take a good pair of headphones and a known, good DAC (not the one in the PC) to those files before I do anything else. Headphones will let you hear all that's there, and a known good DAC gets away from any transmission issues. You may not have a source quality issue....
Frank
fbov is offline  
Sponsored Links
Advertisement
 
post #4 of 37 Old 09-11-2015, 02:28 PM
AVS Forum Addicted Member
 
cel4145's Avatar
 
Join Date: Mar 2009
Location: Grand Rapids, MI
Posts: 16,705
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2668 Post(s)
Liked: 3016
Quote:
Originally Posted by Stereodude View Post
1) 320kbit/sec MP3 files that are encoded by a decent encoder are going to be completely indistinguishable from the CD original to nearly all listeners even on a decent hifi system. However, if you have a dedicated 2 channel listening room with very finely tuned acoustics that can achieve real 3D depth to music the MP3s will not image like the original CD.
+1

Yep. A large majority of people typically find them indistinguishable, even with very good speaker or headphone setups. Download Foobar to your computer and try this test for yourself: http://lifehacker.com/5903625/mp3-or...with-this-test
kitti likes this.

Your questions are answered: Speaker FAQ
Living room HT: Energy RC-50, Ascend Sierra Horizon w/RAAL, Ascend Sierra 1, PSB Imagine XA | PSA V1500 and CHT SS 18.1 | Denon X4200W | modified Dayton SA1000
Other rooms: CBM-170 SE | SVS SB-1000 | Audio-GD NFB-11 | Parasound Zamp | JBL LSR305
Headphone & Portable AKG K7XX | HE-400i | NAD HP50 | Sony MDR-1A | Soundmagic E50 & E80 | X5ii | DX50 | E12
cel4145 is offline  
post #5 of 37 Old 09-12-2015, 04:08 AM
FMW
AVS Forum Special Member
 
FMW's Avatar
 
Join Date: Jun 2007
Posts: 7,480
Mentioned: 13 Post(s)
Tagged: 1 Thread(s)
Quoted: 1654 Post(s)
Liked: 1710
Quote:
Originally Posted by Stereodude View Post
1) 320kbit/sec MP3 files that are encoded by a decent encoder are going to be completely indistinguishable from the CD original to nearly all listeners even on a decent hifi system. However, if you have a dedicated 2 channel listening room with very finely tuned acoustics that can achieve real 3D depth to music the MP3s will not image like the original CD.

2) FLAC files are 100% identical to the CD if ripped correctly. They will sound identical.

There's no magic necessary to send audio files around your house. I'd check that your Plex server isn't recompressing the audio to some low quality / low bitrate MP3 setting, but sending out the original unaltered audio.
People want to hate digital compression for some reason. I'd love to prove to him that the properly made compressed file like these are truly indistinguishable from CD's. We hear what we hear. But we also hear what we expect to hear.
m. zillch and Greenwood Ave like this.
FMW is offline  
post #6 of 37 Old 09-13-2015, 11:18 PM
AVS Forum Special Member
 
PrimeTime's Avatar
 
Join Date: Mar 2003
Location: Lower California
Posts: 3,261
Mentioned: 10 Post(s)
Tagged: 0 Thread(s)
Quoted: 803 Post(s)
Liked: 501
Why is it so important to "prove" to others that your personal opinions and convictions are true?

I have reached my own conclusions regarding perceptual encoding, acoustical treatments and loudspeaker performance/preferences. Others feel differently. If it adds to the discussion and possibly helps others change their opinions (either towards mine, or against), I may offer them from time to time.

If I fail to convince them of my views -- or worse, bring scorn and derision down upon me -- it's no skin off my potato.

In much of this field, we are dealing with subjective entities. Some may invoke "science" or "surveys" as clubs to wield in argument, but true science rarely draws much more than a rough outline here.
PrimeTime is offline  
post #7 of 37 Old 09-14-2015, 12:31 AM
AVS Forum Addicted Member
 
cel4145's Avatar
 
Join Date: Mar 2009
Location: Grand Rapids, MI
Posts: 16,705
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2668 Post(s)
Liked: 3016
Quote:
Originally Posted by PrimeTime View Post
Why is it so important to "prove" to others that your personal opinions and convictions are true?
Because people come here for help, and then you have others post uniformed stuff like this previous post in the thread that steers people down the wrong path:

Quote:
Originally Posted by Speed Daemon View Post
There's no surprise there. Every form of lossy compression, MP3 included, throws away most of the audio information to make a more compact payload. A good codec will pick the least significant parts to delete, which is fine for a telephone conversation or watching the news on TV. But music often has its soul in the subtle parts.
Really? MP3 "throws away most of the audio information?" And the implication here is that the "soul" of the music is lost through compression? You agree with that?

If someone doesn't stand up to the audio myths and legends, they will eventually take over. It's a constant battle.

Your questions are answered: Speaker FAQ
Living room HT: Energy RC-50, Ascend Sierra Horizon w/RAAL, Ascend Sierra 1, PSB Imagine XA | PSA V1500 and CHT SS 18.1 | Denon X4200W | modified Dayton SA1000
Other rooms: CBM-170 SE | SVS SB-1000 | Audio-GD NFB-11 | Parasound Zamp | JBL LSR305
Headphone & Portable AKG K7XX | HE-400i | NAD HP50 | Sony MDR-1A | Soundmagic E50 & E80 | X5ii | DX50 | E12
cel4145 is offline  
post #8 of 37 Old 09-15-2015, 10:32 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
Quote:
Originally Posted by PrimeTime View Post
Why is it so important to "prove" to others that your personal opinions and convictions are true?
Because we're dealing with science here and not people's feelings & emotion. Generally speaking someone is right and someone is wrong.

Quote:
In much of this field, we are dealing with subjective entities. Some may invoke "science" or "surveys" as clubs to wield in argument, but true science rarely draws much more than a rough outline here.
And this is largely what's wrong with much of high end audio today. Too much nonsense and treating customers and potential customers as idiots by not dealing in easily provable facts and reality. Then, the high end community as a whole sits and bemoans that their industry is dying. They've brought it on themselves with the acceptance of things like ceramic cable elevators and other gimmickry nonsense. A total and utter lack of credibility doesn't help sell your products.
Stereodude is offline  
post #9 of 37 Old 09-15-2015, 10:38 AM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
Quote:
Originally Posted by cel4145 View Post
Really? MP3 "throws away most of the audio information?"
Well, from a sheer bitrate standpoint that's true.

By the same standard JPEG and MPEG compression throw away most of the visual information too and at much higher ratios, but in general I don't see people worked up about those lossy compression algorithms like they do with lossy audio compression algorithms.
Speed Daemon likes this.
Stereodude is offline  
post #10 of 37 Old 09-15-2015, 11:23 AM
AVS Forum Addicted Member
 
cel4145's Avatar
 
Join Date: Mar 2009
Location: Grand Rapids, MI
Posts: 16,705
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2668 Post(s)
Liked: 3016
Quote:
Originally Posted by Stereodude View Post
Well, from a sheer bitrate standpoint that's true.
Yeah. But bitrate does not equal audio information. How much actually "audio information" is actually lost? It's a container for audio information. Just because I have a 1 liter bottle with some water in it, that doesn't mean I lose most of the water if I switch to a 16 oz bottle. Depends on how much water was in there.

Not all of the mp3 compression process is lossy. And look at flacs which can easily reduce a wav file down to 50 or 60% of its original size. So 320k? Is that most of the audio information? If the lossless part of the mp3 compression is as efficient as flac, does it?

Your questions are answered: Speaker FAQ
Living room HT: Energy RC-50, Ascend Sierra Horizon w/RAAL, Ascend Sierra 1, PSB Imagine XA | PSA V1500 and CHT SS 18.1 | Denon X4200W | modified Dayton SA1000
Other rooms: CBM-170 SE | SVS SB-1000 | Audio-GD NFB-11 | Parasound Zamp | JBL LSR305
Headphone & Portable AKG K7XX | HE-400i | NAD HP50 | Sony MDR-1A | Soundmagic E50 & E80 | X5ii | DX50 | E12
cel4145 is offline  
post #11 of 37 Old 09-15-2015, 12:14 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
Quote:
Originally Posted by cel4145 View Post
Yeah. But bitrate does not equal audio information. How much actually "audio information" is actually lost? It's a container for audio information. Just because I have a 1 liter bottle with some water in it, that doesn't mean I lose most of the water if I switch to a 16 oz bottle. Depends on how much water was in there.

Not all of the mp3 compression process is lossy. And look at flacs which can easily reduce a wav file down to 50 or 60% of its original size. So 320k? Is that most of the audio information? If the lossless part of the mp3 compression is as efficient as flac, does it?
Oh, I understand that at its heart it is a very complicated question without a simple straightforward answer. Hence my tongue in cheek reply about the bitrate.
Stereodude is offline  
post #12 of 37 Old 09-15-2015, 12:36 PM
AVS Forum Addicted Member
 
cel4145's Avatar
 
Join Date: Mar 2009
Location: Grand Rapids, MI
Posts: 16,705
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2668 Post(s)
Liked: 3016
Quote:
Originally Posted by Stereodude View Post
Oh, I understand that at its heart it is a very complicated question without a simple straightforward answer. Hence my tongue in cheek reply about the bitrate.
I got that

I just thought I'd see if I could get the conversation going about exactly how much audio information is lost

Your questions are answered: Speaker FAQ
Living room HT: Energy RC-50, Ascend Sierra Horizon w/RAAL, Ascend Sierra 1, PSB Imagine XA | PSA V1500 and CHT SS 18.1 | Denon X4200W | modified Dayton SA1000
Other rooms: CBM-170 SE | SVS SB-1000 | Audio-GD NFB-11 | Parasound Zamp | JBL LSR305
Headphone & Portable AKG K7XX | HE-400i | NAD HP50 | Sony MDR-1A | Soundmagic E50 & E80 | X5ii | DX50 | E12
cel4145 is offline  
post #13 of 37 Old 09-15-2015, 08:49 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
Quote:
Originally Posted by Speed Daemon View Post
That's how lossy compression works. Learn the algorithms, do the math and see for yourself. I'll thank you to not ridicule things that you do not know.
*sigh*... If most of the information was lost it wouldn't be acoustically transparent. It seems maybe you're the one who who doesn't understand.
Stereodude is offline  
post #14 of 37 Old 09-15-2015, 09:14 PM
AVS Forum Addicted Member
 
Stereodude's Avatar
 
Join Date: Jan 2002
Location: Detroit Metro Area
Posts: 15,355
Mentioned: 41 Post(s)
Tagged: 0 Thread(s)
Quoted: 4343 Post(s)
Liked: 3082
Quote:
Originally Posted by Speed Daemon View Post
I'm puzzled about why you have gone from making solid, topical remarks to making non sequtur personal attacks. You're the only one here talking about "acoustically transparent" here. Please own your own work.
There's no personal attack. You accuse cel4145 of not knowing what he's talking about when the reverse seems to be the case. I simply pointed that out. What's good for the goose is good for the gander. I also pointed out that most of the information can't be lost if they're transparent (ie: indistinguishable).
Stereodude is offline  
post #15 of 37 Old 09-15-2015, 09:34 PM
AVS Forum Addicted Member
 
cel4145's Avatar
 
Join Date: Mar 2009
Location: Grand Rapids, MI
Posts: 16,705
Mentioned: 35 Post(s)
Tagged: 0 Thread(s)
Quoted: 2668 Post(s)
Liked: 3016
Quote:
Originally Posted by Speed Daemon View Post
That's how lossy compression works. Learn the algorithms, do the math and see for yourself. I'll thank you to not ridicule things that you do not know.
Really? What percentage is lost then when a wav file is converted to 320k mp3? Why don't you tell us?

I suspect you would have already if you actually knew

Your questions are answered: Speaker FAQ
Living room HT: Energy RC-50, Ascend Sierra Horizon w/RAAL, Ascend Sierra 1, PSB Imagine XA | PSA V1500 and CHT SS 18.1 | Denon X4200W | modified Dayton SA1000
Other rooms: CBM-170 SE | SVS SB-1000 | Audio-GD NFB-11 | Parasound Zamp | JBL LSR305
Headphone & Portable AKG K7XX | HE-400i | NAD HP50 | Sony MDR-1A | Soundmagic E50 & E80 | X5ii | DX50 | E12
cel4145 is offline  
post #16 of 37 Old 09-15-2015, 09:50 PM
Senior Member
 
bouf0010's Avatar
 
Join Date: Oct 2014
Location: Ottawa, Canada
Posts: 324
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 144 Post(s)
Liked: 70
http://ryanmaguiremusic.com/theghostinthemp3.html

Sent from my SM-G925W8 using Tapatalk
bouf0010 is offline  
post #17 of 37 Old 09-15-2015, 10:20 PM
AVS Forum Special Member
 
A9X-308's Avatar
 
Join Date: Mar 2008
Location: Australia
Posts: 8,331
Mentioned: 29 Post(s)
Tagged: 0 Thread(s)
Quoted: 1796 Post(s)
Liked: 1488
Quote:
Originally Posted by bouf0010 View Post
From the article: "Listening tests, primarily designed by and for western-european men, and using the music they liked, were used to refine the encoder."

White patriarchal heteronormativity strikes again.





The article was interesting, but rather than listening to the removed content, what is far more relevant to the average person, is can they tell the difference between 16/44 and 320kbps when listening to actual musical material? For most people, most of the time, that's no.
A9X-308 is offline  
post #18 of 37 Old 09-16-2015, 12:30 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by A9X-308 View Post
The article was interesting, but rather than listening to the removed content, what is far more relevant to the average person, is can they tell the difference between 16/44 and 320kbps when listening to actual musical material? For most people, most of the time, that's no.
This is perhaps the more balanced view. Lossy compression is not a set of fixed conditions because it involves perceptual coding that responds to changes in the input signal. It is, therefore, variable by nature. That means that under some conditions it's going to leave tracks, under others, none at all, and frankly everywhere in between. It's simply not a black/white question, not at 320kpbs or most other bit rates. However, statistically, if it's acceptable or even transparent to most listeners most of the time, you have a win most of the time. Can you say it's always 100% transparent? Nope, but most of the time it doesn't matter. But it's not a yes/no, black/white, good/bad answer.

There are some similarities in visual compression, but not always. up the thread a ways Stereodude made a parallel with visual compression, "By the same standard JPEG and MPEG compression throw away most of the visual information too and at much higher ratios, but in general I don't see people worked up about those lossy compression algorithms like they do with lossy audio compression algorithms." With video compression, most people never actually attempt compression of an uncompressed source, so you don't hear the griping much. But if you actually do that, you're in for a real rough ride.

You don't hear people getting upset as much about video compression because a lot of the compression tweaks have been done for them, optimizing the results, without access to the uncompressed original. However, as someone who has authored just a few DVDs, where visual compression must take place, we often go to extreme lengths to adjust things about the compressor to trick it into doing a more transparent job. Dropping markers into the material at critical moments that tweak the compressor to work harder on that moment, carefully choosing maximum and minimum bit rates then manually adjusting them scene by scene, various types of pre-processing. Then the compressor is allowed make two or more passes through the material to fine-tune itself. All of that lets us create compressed content that looks largely transparent...so long as you don't have the uncompressed original to compare to, which most people don't have. But if you had the original, the degradation is almost always detectable at some point.

With still images, cameras have all be pre-set for optimal image compression, and only offer the user a few settings, each of which has already been optimized for its target resolution and file size for that camera. So the hard work has been done. But, if you want to hear people gripe about compression, pop into the pro photo world. No .jpgs being shot there at all. None. All camera-raw (uncompressed). Why? Because .jpg compression beats up your image. And for folk wiring in a world of grey scales, they're pretty black and white about that. For everyone else, the "most/most" concept works just fine, and they only shoot .jpg images.

When you run audio through an .mp3 coder you have only a single bit rate to select for the entire file. You can't manually tweak it on the fly. You can choose either fixed or variable rate, but how variable it is and how and when it chooses vary are pre-set and not optimizable. There may be a tiny fist full of other tweaks, but it's mostly a blanket codec that you can't really do much with, you just take what you get for a given bit rate, there's not a lot you can do. And people rip their own uncompressed originals all the time, it's an easy comparison to do. If you don't like what you get, you're only real option is to up your bit rate or find a better codec.

That's why I think you don't hear as much moaning about visual codecs, at least not in an audio forum.

Just because...I used to squash things into .mp3/320kbps files. Unfortunately, I got exposed to the "mp3 swish" decades before the .mp3 file was popular, and once you've heard it and recognized it, you become sensitized and hear it everywhere quite easily, even at high bit rates. So, even in my 320k files I could often hear the swish and warble. 128kbps is just agony. That got me to AAC, 320kbps, which is mostly artifact-free. And for me, that hit the most-of-the-time mark. I also do lossless (on the computer HDD), and that's where I've stayed, except for portable devices with tiny little brains that must have squashed files or they just won't hold enough media.
jaddie is offline  
post #19 of 37 Old 09-16-2015, 09:53 AM
Advanced Member
 
James Freeman's Avatar
 
Join Date: Sep 2013
Posts: 757
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 517 Post(s)
Liked: 348
I love these threads.

If you can't hear past 16kHz I would argue that you will not distinguish 128kbps MP3 to a CD, let alone 192kHz HiFi recordings.
I came to a conclusion that my MP3 collection is perfectly adequate for my enjoyment even though I'm in my late 20s and still hear 19kHz.

I own active studio monitors with ribbon tweeters and a HiFi Single-Ended tube amp, and a nice low distortion reference (Class AB) solid state amplifier with good hifi speakers (Polk RTi A3).
This setup covers all my HiFi needs and beyond, and I'm still perfectly content with my 128kbps MP3s.

Cheers.
James Freeman is offline  
post #20 of 37 Old 09-16-2015, 10:04 AM
AVS Forum Special Member
 
fbov's Avatar
 
Join Date: Nov 2006
Location: Bushnell's Basin, NY
Posts: 1,632
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 328 Post(s)
Liked: 196
Quote:
Originally Posted by jaddie View Post
... Unfortunately, I got exposed to the "mp3 swish" decades before the .mp3 file was popular, and once you've heard it and recognized it, you become sensitized and hear it everywhere quite easily, even at high bit rates....
Quote:
Originally Posted by James Freeman View Post
I love these threads.

If you can't hear past 16kHz I would argue that you will not distinguish 128kbps MP3 to a CD, let alone 192kHz HiFi recordings....
While you're right from a sampling standpoint, and MP3 has certainly found a niche, that's all irrelevant if there's a "tell" in the compression process. I suspect the swish is well below 16KHz.

Frank
fbov is offline  
post #21 of 37 Old 09-16-2015, 10:12 AM
Advanced Member
 
James Freeman's Avatar
 
Join Date: Sep 2013
Posts: 757
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 517 Post(s)
Liked: 348
Define swish please.
I will post some left overs of the compression for you to hear what mp3 throws out at various bit rates, using a phase flip trick (nulling).
James Freeman is offline  
post #22 of 37 Old 09-16-2015, 10:57 AM
AVS Forum Special Member
 
jim19611961's Avatar
 
Join Date: Mar 2011
Location: Texas
Posts: 1,829
Mentioned: 13 Post(s)
Tagged: 0 Thread(s)
Quoted: 451 Post(s)
Liked: 158
Quote:
Originally Posted by James Freeman View Post

If you can't hear past 16kHz I would argue that you will not distinguish 128kbps MP3 to a CD, let alone 192kHz HiFi recordings.
I came to a conclusion that my MP3 collection is perfectly adequate for my enjoyment even though I'm in my late 20s and still hear 19kHz.
On my system with my ears, there is a WORLD of difference between 128Kbps mp3's and CD, and my ears top out at about 15K. And an audible one for 320k and CD.

My Room
My Music
Rega - Apollo, Rega - DAC, Goldpoint Passive, Classe CA-100, Crown XLS 1002, miniDSP
Jenzen Next
jim19611961 is offline  
post #23 of 37 Old 09-16-2015, 11:05 AM
Advanced Member
 
James Freeman's Avatar
 
Join Date: Sep 2013
Posts: 757
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 517 Post(s)
Liked: 348
Here is a 17 second samples of Michael Jackson's Billie Jean in various formats to demonstrate what exactly MP3 throws out from the original:

http://www.mediafire.com/download/vb...tv/Michael.zip

Original Flac.wav
320.mp3
320 thrown.wav
192.mp3
192 thrown.wav
128.mp3
128 thrown.wav

If you combine the "128.mp3" + "128 thrown.wav" in your DAW you will get exactly the Original Flac back and it will null to silence if phase flipped with the original.

I don't know what the swish nonsense is all about, but this test allows you to listen to how MP3 works and what the compression algorithm throws away based masking of human hearing.
I was amazed how much the 128kbps throws out.. almost a listenable copy of the song, but then I realized I'm only a human and frequency masking applies on me too.

It was fun forcing myself to hear the thrown part in the 128.mp3, but I couldn't...

Hope this helps.

Last edited by James Freeman; 09-16-2015 at 11:19 AM.
James Freeman is offline  
post #24 of 37 Old 09-16-2015, 11:06 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by fbov View Post
While you're right from a sampling standpoint, and MP3 has certainly found a niche, that's all irrelevant if there's a "tell" in the compression process. I suspect the swish is well below 16KHz.

Frank
I hate to describe it in detail because more people with then hear it, and it will ruin the experience. The "swish" is largely upper mid-band but highly variable and program dependant.

I can even hear it on some radio commercials, most of which have been distributed to stations by downloadable, low rate .mp3 for almost 20 years. There used to be a dedicated distribution system using a purpose-built PC/sound card/display in a rack-mounted unit...forgot the name...probably no longer used. I was Chief Engineer at several stations in another life.

The artifact doesn't require a high performance system to hear, a car radio is just fine. The artifact is less a higher bit rates, but once trained to hear it, you hear it everywhere. Negative conditioning. It's a curse. First heard it in the mid 1980s when we tried using various codecs and bit rates for audio transport over ISDN.
jaddie is offline  
post #25 of 37 Old 09-16-2015, 11:07 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by James Freeman View Post
Define swish please.
I will post some left overs of the compression for you to hear what mp3 throws out at various bit rates, using a phase flip trick (nulling).
I don't think I'd be doing anyone a service. I view hearing it as a curse, because it's everywhere.
jaddie is offline  
post #26 of 37 Old 09-16-2015, 11:08 AM
Advanced Member
 
James Freeman's Avatar
 
Join Date: Sep 2013
Posts: 757
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 517 Post(s)
Liked: 348
Quote:
Originally Posted by jim19611961 View Post
On my system with my ears, there is a WORLD of difference between 128Kbps mp3's and CD, and my ears top out at about 15K. And an audible one for 320k and CD.
Please...

Quote:
Originally Posted by jaddie
I don't think I'd be doing anyone a service. I view hearing it as a curse.
It is done.
Download and listen.
James Freeman is offline  
post #27 of 37 Old 09-16-2015, 11:21 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by James Freeman View Post
Please...


It is done.
Download and listen.
Creating examples and not describing what to listen for isn't exactly "It is done".

Don't have the time to play with the files right now, but I will later.

"Swish" is my on name for the artifact, but it's a descriptive term. Sorry if you think it's "nonsense", it isn't to me. I played the first 5 seconds of your 128 file on my laptop with tiny speakers in a noisy hotel lobby, hear it immediately, no question.

Yeah, I know...and I actually am a hard-core ABX guy. Please don't beat me up over a very casual quickly formed non-scientific subjective opinion.
jaddie is offline  
post #28 of 37 Old 09-16-2015, 11:25 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by James Freeman View Post
Please...


It is done.
Download and listen.
Just another few seconds to listen, I really have to run...

The "thrown" files don't illustrate the artifact I'm talking about at all, the do contain the less audible results of the coded. Sorry. The content of them actually will mask the artifact.

More later, sorry, business calls.
jaddie is offline  
post #29 of 37 Old 09-16-2015, 11:56 AM
Advanced Member
 
James Freeman's Avatar
 
Join Date: Sep 2013
Posts: 757
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 517 Post(s)
Liked: 348
Here is the same thing with classical music (Bach):
http://www.mediafire.com/download/c4...h0aqx/Bach.zip

Original Flac.wav
320.mp3
320 thrown.wav
128.mp3
128 thrown.wav

Far more complex and hard on the compression algorithm, lots of high frequency violins.
James Freeman is offline  
post #30 of 37 Old 09-17-2015, 01:59 AM
AVS Forum Special Member
 
Join Date: Jan 2007
Posts: 2,630
Mentioned: 14 Post(s)
Tagged: 0 Thread(s)
Quoted: 1331 Post(s)
Liked: 817
Quote:
Originally Posted by James Freeman View Post
Here is the same thing with classical music (Bach):
http://www.mediafire.com/download/c4...h0aqx/Bach.zip

Original Flac.wav
320.mp3
320 thrown.wav
128.mp3
128 thrown.wav

Far more complex and hard on the compression algorithm, lots of high frequency violins.
Looking at the files in VLC, the codec used comes out Mpeg 1/2 mpga, not mp3.

The codec used has an obvious low pass filter, not all do (Fraunhofer doesn't, for example).

Might be good to compare apples and apples...

There are several different codecs around, they don't all perform identically. There are several versions of each, later versions might be assumed to be improvements.

With all those variables, you might well produce some really transparent .mp3 files (though VLC says you didn't), where someone else might produce some very non-transparent .mp3 files, using the same bit rate but different codecs and settings.

I don't think it makes sense to try to prove .mp3 at various bit rates are audibly transparent or not. That's been beat to death.

The .mp3 artifact I have referred to as "swish" is a mid-band artifact, and may be the result misapplied codecs or settings, and may be avoided with careful codec choice and setup. It's a coding error, where the encoder has either misapplied the psychoacoustic model, or been tricked by program material, or might be just a less capable encoder. The best way to describe it is like someone switching a band pass filter on and off quickly, modulated by audio. It's obvious at low bit rates, but is easily heard at 128k too. The exact cause? I don't know, and don't care. This mattered a lot more when we didn't have the storage we have now. But I find it very easy to hear, recognize it very quickly, and hear it in an alarming number of situations, considering the storage and transmission bandwidth we now have. In fact, driving home to day, listening to an FM station, they played a spot with the "swish" all over it.

I've set up several broadcast audio systems for audio storage/retrieval, and had to determine how we would encode various program elements. We did it with a controlled ABX test (I actually own a hardware ABX comparator), comparing the same source material uncompressed, and with various rates of mp3. We determined that 128k, CBR, Stereo, 44.1KHz Fraunhofer mp3 was differentiated from the original quite often, 320Kbps (other parameters held constant) was not reliably differentiated from the original. We also determined that Joint Stereo was easily differentiated, at any bit rate. We did not test VBR because it created other issues with time readouts. We did not test other encoders (LAME, Blade, Xing, etc.). We also did a much more cursory test of mp3 audio at several rates passed through a second, non-mp3 codec, as it might for transmission systems, and found some odd results. Sometimes the mp3 actually sounded better than the uncompressed through the second codec.

So we can beat this up all day, but with storage now what it is, I don't know why anyone would use anything less than 320Kbps, CBR, Stereo, Fraunhofer (beats LAME for CBR), or if the option is available, AAC at 320K, which is absolutely transparent.

FWIW, the classical samples were much cleaner at 128 than Michael, on which I heard the artifact easily and within a few seconds.
jaddie is offline  
Sponsored Links
Advertisement
 
Reply Audio Theory, Setup, and Chat

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off