Can you hear the differences between 128/192/VBR and 320kpbs?

[DeoxySynchro]

For me, I really couldn't hear the differences clearly between 128/192/VBR and 320KBPS mp3's.

Let's take Chrono Symphonic for example. There are 2 version for this OST, which are FLAC and MP3. When I listen to FLAC and MP3 (VBR), I really couldn't tell much the differences, while FLAC took up almost 10 MB each file and MP3 only use about 3MB for each file.

Does it requires a good soundcard like CREATIVE SOUND BLASTER LIVE in order to hear those differences clearly? For now I am using the old soundcard by Creative, couldn't think of the name though... Btw, I hear that if someone could hear the difference, that guy must have a good pair of ears, wonder is that true?

Most mp3's I get by downloading/leeching from the net such as anime and game OST are usually encoded with LAME on 128/192 and VBR. Only few singles are encoded with 320kbps.

[Vie]

I can hear the difference between a 128 and a 192 encoded MP3, quite easily in fact. I dont have that great of a soundcard, just a generic 5.1 one. Its an "Aureon 5.1 Fun Wave" if you want to know, bad name I know, it was cheap and works just fine.

As to 192 compared to 320, I can again hear the difference, but its harder to do. For preference I go for 256 LAME vbr when I have to use MP3, I cant normaly hear the difference between a 256 and a 320 and its a slightly smaller filesize.. And Ogg Vorbis at Q8 the rest of the time, I prefer using Vorbis for a number of reasons.

The main one is its decoder is slightly easyer on my Palms battery life than the MP3 decoder. No idea why, but I get more play time with Ogg than with MP3.

[DeoxySynchro]

Good for you. Frankly, after listening to so many 128/192/VBR and 320kbps song, I still couldn't hear the differences clearly, especially for vocal singles.

Some question:

1) Is the file size of OGG the same as MP3?
2) Is OGG loseless format like FLAC?

[Vie]

Ogg FLAC is lossless, Ogg Vorbis is not.
Ogg tends to be a smaller filesize than MP3 at the same vbr bitrate.

[zootm]

Originally posted by Vie
Ogg tends to be a smaller filesize than MP3 at the same vbr bitrate.

No, two files of the same length with the same bitrate will (well, should; with VBR it's kind of an estimate) always be the same size. Vorbis is better quality than MP3 at the same bitrate, though, so you can encode it at a lower bitrate to get a file of the same quality, which is smaller.

[Vie]

No, I've found that Vorbis tends to be smaller at the same bitrate myself. Not much smaller, but a few kb can make all the difference.

[gaekwad2]

That's most likely due to rounding of different ways of calculating the average bitrate though.

As for the original question, depending on the encoder and settings 128 can be anything between bloody awful and transparent most of the time. 192 is generally transparent to me.

[germain47]

I think we tend to forget the hardware we currently use for playback will likely improve in quality over time. When I finally sprung for some high end headphones, the difference in quality compares to a bright sunny day to impending twilight. I almost cringe listening to the 128 CBR MP3 files I downloaded years ago. While VBR at about 128kbs (V 5)is noticably better, I still have little trouble picking these out. Most often a VBR at about 192kbs(V 2) reaches transparency for me, though I encode at a little bit higher rate just for a margin of saftey. I'd rather give up file size than quality.

I rarely encode with vorbis for compatibility reasons only. I find transparency can be reached at a somewhat lower bitrate. I have read somewhere that the artifacts introduced by vorbis tend to be more pleasing to the ear than the artifacts mp3's introduce. Maybe that's why I feel no need to re-encode my Beatle's collection which is at ~112kbs vorbis.

[DeoxySynchro]

It's been troubeling me about those TRANSPARENCY jiggy-maggy thing. I really couldn't understand the meaning of it. Can someone explain it?

[gaekwad2]

Transparent means being unable to hear a difference between the compressed (in this case) version and the original.

[germain47]

Transparncy simply means that when you compare the orginal CD source to the compressed lossy (mp3,vorbis,etc) file, you can not detect any difference in sound quality, ie. you could not pick one from another in a blind test.

When files are compressed using a lossy format, some of the audio information present in the orginal source is forever removed. Mostly this is audio you wouldn't be able to hear anyway. Like someone quietly talking to a companion in a movie theatre four rows away during a loud section of the movie. Also, lossy compression uses lowpass filters to remove audio above a given frequency level. These very high frequencies are only marginally audiable to most people. Finally, for reasons compicated to explain, sounds that were not orginally present (artifacts) are introduced. The more one forces the encoder to compress the source, the more audio is removed from the source, the lowpass filter is set lower and wider band of high frequencies are removed and more artifacts are introduced.

[edit] Hi Gaekwad! Guess I'm too slow.

[gaekwad2]

Artifacts are created because the encoder doesn't directly remove information (except for lowpass and things like intensity stereo) but choose a space saving but less accurate method of representing the original signal.

[DeoxySynchro]

Woah! That's a bunch of useful information you got there, germain and gaekwad2. Guess I'm started to understand those terms used throughout mp3 encoding. Thanks a lot.

[rockouthippie]

In a blind taste test, I can tell a 128k MP3 from the original. Really, my ear can tell whether I am listening to an MP3 at 128k.

MP3 @ 128kbps is 85% accurate, so it's not surprising that my ear would hear the difference.

At 160 kbps CBR, I might be able to tell.

At 160 VBR, I doubt it unless it was some synth music.

[deeder7001]

160kbps is the worst for me. The highs are crappy sounding at that bitrate to me. I have all my CDs at 320 alt-preset insane on my computer. A high bitrate vbr and alt-preset insane I can't really tell a big difference. Between 128 and 160 I can tell a difference clearly(for most of my music at least).

[ScorLibran]

Can you hear the differences between 128/192/VBR and 320kpbs?

The only way to know is to use double-blind, hidden-reference differential analysis using an ABX comparator.

Or you can reference the results of listening tests other people have completed. A lot of thorough information can be found that way.

Anyone making their own claims of hearing differences between different types of audio encoding are full of placebo until proven otherwise. In other words, they're speculating.

[gaekwad2]

It's hard to ABX if you don't have the originals though.

(For the record, I managed to ABX some artifacts in Lame -V5 (~135kbps) encodes of 'normal music', but none at -V4 (~160).)

[ScorLibran]

You don't need originals to ABX different encoding methods against each other. Call 320 CBR the reference, and go from there.

[rockouthippie]

I just use flac.....

Hard disks are big enough these days. And re-encoding to whatever is fast enough to load a MP3 player up with tunz.

[ScorLibran]

Originally posted by rockouthippie
Hard disks are big enough these days.

True. I was preaching that last year, when hard drives were also big enough.

The only reason I'm currently transcoding all my music to MP3, with the intent of deleting all the FLAC originals, is because I want to fit all my music on my notebook's little 100 GB hard drive.

Right now I've got almost 300 GB in FLAC encodings. And I'm not keeping lossless backups online because the CDs are the backups (yeah, I'm weird, I've got CDs for all my music), and I need my big network drives for my business.

[JFASI]

I'm stuck with my mac's internal speaker, in which case I can't tell the difference between 32 bit and 192 bit.

[DeoxySynchro]

I just encode my mp3's with both 64kbps | 22 hz and 320kbps | 44 hz. The differences is clear. However, when I encode with 128kbps and 320kbps, I still couldn't tell the differences much.

[deeder7001]

The differences between 128kbps and 320kbps aren't that great unless you have a good stereo system setup and you know what to listen for(in my experience at least).

[ScorLibran]

Placebo is a powerful effect, even with a good stereo system.

[psygoa]

Originally posted by DeoxySynchro
I just encode my mp3's with both 64kbps | 22 hz and 320kbps | 44 hz. The differences is clear. However, when I encode with 128kbps and 320kbps, I still couldn't tell the differences much.

Very true. If you have the usual PC speaker setup it won't matter since the information these mp3s are leaving out in the lower bitrates won't be played anyway. If you get the chance listen to a few mp3s at 128 on a really good system like a pair of studio monitors. The difference is ridiculous. 320s sound different to me than the original pure wave file on my monitors. It sounds colder in the highs and muddy in the lows since that is where the codec operates.

[ScorLibran]

Originally posted by ScorLibran
Placebo is a powerful effect, even with a good stereo system.

[gaekwad2]

I sometimes wish audiophile posers would at least read up on artifact descriptions, it's far too easy to spot them this way.

[ScorLibran]

They always "know what they hear", and are so sure about it. I was too until I started running double-blind listening tests several years ago. They give you a great respect for the placebo effect.

[shakey_snake]

My curiousity from browsing this thread has led be to this page, but I have decided I'd probably like myself better if I didn't listen to any of those files.

[ScorLibran]

Darrell's got some very strange stuff in there, for sure.

[Warrior of the Light]

I bet everyone can hear the difference between 128 and 192 kbps MP3. 192 to 320 depends on the kind of music you're listening to.. Simple computer generated trance shit sounds more like the original than a recording of an entire orchestra.

(using OGG Vorbis at Quality 8 myself)

[gaekwad2]

OK then,

http://www.badongo.com/file/1919930 (~5MB)

All samples were encoded using Lame 3.97.
Reference is -V2 --vbr-new (usually produces around 200kbps).
Test samples are
-V5 --vbr-new (usually produces around 130kbps)
128kbps ABR
128kbps CBR
128kbps CBR Stereo (others are Joint Stereo)
96kbps CBR

See the included readme on how to take the test.

[rockouthippie]

Originally posted by ScorLibran
(yeah, I'm weird, I've got CDs for all my music)

I suffer from the same weirdness

(using OGG Vorbis at Quality 8 myself)

That is superior to MP3 by a lot.

[ScorLibran]

Originally posted by rockouthippie
That is superior to MP3 by a lot.

Actually it's worse. Anything at or above perceptual transparency is equal in sound quality to anything else at or above perceptual transparency. In other words, "transparency is transparency".

So when comparing two formats and/or encoder settings, both at a level of perceptual transparency, what matters is efficiency.

Vorbis @ -q8 = about 256kbps

LAME MP3 @ --alt-preset standard = about 210 kbps

With Vorbis -q8 you're using some 20% more disk space for the same sound quality as LAME MP3 using the --aps preset.

Vorbis @ -q7 has been shown to also be perceptually transparent, and uses about the same bitrate and disk capacity. Equal efficiency.

For many people's ears, Vorbis @ -q6 may also be be transparent, and if so, it would provide some 10% higher efficiency.

So Vorbis does have the potential to be more efficient than MP3, but not at -q8.

[gaekwad2]

-q6 is generally seen as the aps equivalent (at least since aoTuVb2/Vorbis 1.1).

[Rocker]

depending on the music style, q5 can also be quite transparent.

aps sounds like ass with trance music, when compared to vorbis at q6.

thats what I find anyway. but my ipod doesn't support vorbis doh.

[gaekwad2]

Depending on sensibility/training even q2 can be transparent.

[germain47]

Orginally posted by ScorLibran
Anything at or above perceptual transparency is equal in sound quality to anything else at or above perceptual transparency. In other words, "transparency is transparency".

The key word here is perceptual. What each of us perceives is going to vary. What the audiophile is able to perceive on his $1000+ headphones will be different than what another hears through cheap computer speakers. That's why when possible I archive to disc as flac files or download files somewhat above my perceptual transparency. Who knows, I might grow a wild hair and spring for top end equipment somewhere down the line.

[ScorLibran]

Originally posted by germain47
What the audiophile is able to perceive on his $1000+ headphones will be different than what another hears through cheap computer speakers.

When the audiophile proves his perception with a double-blind listening test, he'll be perceived as something other than foolish.

Until then, he'll only be an expert on hemorrhaging money. And I'll have a $15,000 signal purifier to sell him.

[germain47]

I think you miss the point. When one upgrades from cheap equipment one will notice an improvement in their audio experience. If one downloads or rips to the low end of their perceptual transparency on their cheap stuff, they may be sorely disappointed if and when they upgrade equipment.

Been there, done that. Didn't need an ABX test to tell me I screwed up. Eventually I climbed far enough up the learning curve to do my own tests, but I don't feel the need to bring that smug arrogance over to this forum.