Anybody Out There Who Hears a Difference Between 320 kbps MP3 and Red Book CD? What Differences Do You Hear?

[2M2B]

So mp3 encodes 16kHz and above as one frequency band. And this band has no scale factor that can be changed independently (ref1, ref2), limiting mp3 format's ability to accurately encode this band.

LAME MP3 has a -Y switch which tells the encoder to ignore/cut the 16KHz+ band if causes bloat or wasted bits. With V2 + -Y tracks that are 240kbps for no reason instead be 175 ~ 190kbps.

There have been "home grown" projects that tried to address these fundamental issues, one of the earliest being MP+ (then MPC, now Musepak). It was much more transparent at ~200kbps because it could actually allocate those bits to short transients and high frequencies. But people's HDDs were still small, the algorithms were still patented, and most people outside the enthusiast circles wanted more songs per megabyte, not less (anyone remember mp3pro?)

You forgot Musepack --Standard(170kbps) which despite being a pure subband codec holds fine to AAC/Vorbis/Opus. I have no idea how FHG thought MP3 would be transparent at 128 ~ 175kbps with it issues, Then they added salt to the wound by capping decoders to 320kbps despite MP3 max bitrate was 640kbps?.

That what bugged me about VBR in MP3 what the point if It locks to 320kbps, If a sections or a whole sample could be transparent at 350 ~ 500kbps like in other lossy codecs.

Edit: Some are forgetting that AAC/Vorbis/Opus can reach 512kbps even if at 192kbps VBR, If it is something very complex. While Musepack can go much higher like 1.4mbit if needed, I've had few albums with sections to whole song reach 540 ~ 790kbps even with the --Standard setting.

 

[HRTF_Enthusiast]

And when people tell me they can clearly hear a difference from track to track without too much effort, that just proves they're full of BS.

For this one song, the only effortless part was the beginning with mp3. Other sections of the song were doable as well, but much harder. It's a completely different level with AAC and Vorbis though.

 

[danadam]

But Im sure many enthusiasts use bluetooth AND lossy material and dont realize how bad that combination is.

If they don't realize it then it can't be that bad

 

[HRTF_Enthusiast]

The beginning of this song is also ravaged by 320 kbps mp3 (LAME encoded). On MP3 it's like a "poof" while on lossless it's a "slam."

MediaFire

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

 

[escape2]

The beginning of this song is also ravaged by 320 kbps mp3 (LAME encoded). On MP3 it's like a "poof" while on lossless it's a "slam."

I am glad my old ears aren't sensitive enough to tell the difference, so I don't have to worry about it.

For those who can tell the difference, by all means stay away from MP3.

foo_abx 2.0.6d report
foobar2000 v1.6.8
2021-12-17 10:24:14

File A: taishi flac.wav
SHA1: 2997777d574254b6f897f4acc4f98d036bdc9e02
File B: taishi mp3.wav
SHA1: ddabac4544bc4f1d6dd69c839d200a0323d4a494

Output:
WASAPI (event) : LOXJIE D30 (2- LOXJIE AUDIO), 32-bit
Crossfading: NO

10:24:14 : Test started.
10:29:05 : 00/01
10:29:22 : 00/02
10:29:44 : 01/03
10:30:00 : 02/04
10:30:19 : 03/05
10:30:29 : 03/06
10:30:42 : 04/07
10:31:03 : 04/08
10:31:14 : 05/09
10:31:27 : 06/10
10:31:38 : 07/11
10:31:51 : 07/12
10:42:11 : 07/13
10:42:28 : 07/14
10:42:39 : 07/15
10:42:48 : 07/16
10:42:48 : Test finished.

 ---------- 
Total: 7/16
p-value: 0.7728 (77.28%)

 -- signature -- 
94a3d7d06546699afcff2b98278344c097f4b396

 

[LeftCoastTim]

For this one song, the only effortless part was the beginning with mp3. Other sections of the song were doable as well, but much harder. It's a completely different level with AAC and Vorbis though.

This is why I don't worry about lossless streaming. Spotify (vorbis/aac) and Apple music (aac) is good enough for all cases, and even if I can abx a difference between them and lossless, it reduces my enjoyment of the music by zero amounts.

But people do love "purity" though. It's like a cleansing ritual.

 

[2M2B]

I must be going mad Noticed quite a few quirks with few encoders/codecs:

QAAC : Struggles on ambient at 96 ~ 192kbps.

Nero AAC : Same with QAAC but fails at 240kbps on Grimmrobe demos, Custom settings it transparent at 160 ~ 224kbps despite being old.

FHG AAC: The best out the bunch at 192kbps.

Exhale(xHE AAC): Painfully slow, Worse than above three at 128 ~ 192kbps on ambient.

Vorbis : Worse than MP3/AAC, Too offended to go on further.

Opus: Fantastic at 128kbps pretty much transparent.

Musepack: Same as Opus at 175kbps.

MP3: Suffers from pre echo at any bit rate, very outdated It needs 256 ~ 320kbps on samples that newer codecs are transparent on at 96 ~ 176kbps(assuming good encoder).

Helix MP3: No idea what the dev's did but It pretty much transparent at 192kbps rarely even hear pre echo artifact's. Hard to believe that MP3 can sound this good despite It limitions.

 

[HRTF_Enthusiast]

Vorbis : Worse than MP3/AAC, Too offended to go on further.

320 Vorbis is worse than 320 kbps mp3 for you? Did you ABX this?

 

[LeftCoastTim]

Opus: Fantastic at 128kbps pretty much transparent.

Opus is amazing, isn't it? I can barely tell the difference on a difficult hard track at 160kbps (90% correct on ABX) . Vorbis at same sound quality is at 192kbps. And the difference is not at all objectionable.

If I could stream and purchase tracks at 192kbps opus, I'd be happy.

 

[2M2B]

Opus is amazing, isn't it? I can barely tell the difference on a difficult hard track at 160kbps (90% correct on ABX) . Vorbis at same sound quality is at 192kbps. And the difference is not at all objectionable.

If I could stream and purchase tracks at 192kbps opus, I'd be happy.

Same, I would totally use It If it had way more support. AAC(QAAC/Nero AAC combo) at 160kbps just as good as Opus. Meanwhile MP3 only good 90% of the time at 256kbps with LAME V0.

 

[freemansteve]

Obviously I'm old and deaf (too many live gigs?) among other things, but I followed some of the above links and someone said that Kraftwerk's The Man Machine was hard to rip to MP3 and "obviously sounded worse".

So, I just ripped my 2009 version from FLAC to MP3 (using LAME 3.10, 320 kb/s CBR). I tried both versions by queuing them up and switching between them, using IEMs, normal headphones, and speakers. Not very scientific I agree, but I couldn't here any difference except that volume levels were slightly different - one has MP3Gain set at 93dB and the other has ReplayGain set so no surprise. Certainly, I heard nothing to make me say "eek, horrid MP3!". The MP3 was easily as good as when I saw them live, but obviously different (as was the FLAC). I cannot try my very original CD version, as I am without a player at the moment, bar PC ones I use for ripping - it will sound different anyway to the 2009 version, so no point.

I am not disputing others can hear a difference, but is it necessarily unpleasant, or unlistenable to? In any case, it strikes me that Kraftwerk synthesizes everything (bar occasional vocal) so I'm not sure it's the right album to compare lossless versus lossy.

At last - there is one benefit to being "past it"!!

PS I need MP3 versions of music for my car which cannot play FLACs off my 128G USB stick, which is annoying, but that's another story....

 

[Julf]

PS I need MP3 versions of music for my car which cannot play FLACs off my 128G USB stick, which is annoying, but that's another story....

Same here, but it is really not an issue -sound quality will never be perfect in that car.

 

[SIY]

Same here, but it is really not an issue -sound quality will never be perfect in that car.

Gordon Holt: “High fidelity in a car is like Arpege in a glue factory.”

 

[freemansteve]

Same here, but it is really not an issue -sound quality will never be perfect in that car.

Definitely - always is dreadful in a car, no matter what you spend on the system, but some stirring beats is good on a long motorway.... Better to spend it on a quieter car than a car system!

 

[freemansteve]

Gordon Holt: “High fidelity in a car is like Arpege in a glue factory.”

I had to look that one up!

The Crushing Disappointment of L’Arpège

Is the big-game-trophy model of destination dining really worth it anymore?

www.eater.com

 

[SIY]

I had to look that one up!

The Crushing Disappointment of L’Arpège

Is the big-game-trophy model of destination dining really worth it anymore?

www.eater.com

I think Gordon was referring to Arpege the perfume. Nonetheless, valid.

 

[ahofer]

Obviously I'm old and deaf (too many live gigs?) among other things, but I followed some of the above links and someone said that Kraftwerk's The Man Machine was hard to rip to MP3 and "obviously sounded worse".

So, I just ripped my 2009 version from FLAC to MP3 (using LAME 3.10, 320 kb/s CBR). I tried both versions by queuing them up and switching between them, using IEMs, normal headphones, and speakers. Not very scientific I agree, but I couldn't here any difference except that volume levels were slightly different - one has MP3Gain set at 93dB and the other has ReplayGain set so no surprise. Certainly, I heard nothing to make me say "eek, horrid MP3!". The MP3 was easily as good as when I saw them live, but obviously different (as was the FLAC). I cannot try my very original CD version, as I am without a player at the moment, bar PC ones I use for ripping - it will sound different anyway to the 2009 version, so no point.

I am not disputing others can hear a difference, but is it necessarily unpleasant, or unlistenable to? In any case, it strikes me that Kraftwerk synthesizes everything (bar occasional vocal) so I'm not sure it's the right album to compare lossless versus lossy.

At last - there is one benefit to being "past it"!!

PS I need MP3 versions of music for my car which cannot play FLACs off my 128G USB stick, which is annoying, but that's another story....

It takes training and listening for very specific things (in my case cymbals and/or strings) to hear a difference between the highest resolution MP3 and true lossless. Even then, I don't do better than chance with pop music, and a little better with classical. While I'm 58, I believe the same is true of much younger folks.

Prove your audiophile skills with this blind test on lossless music

Audio enthusiasts often pride themselves on their ability to discern the tiniest nuances of a recording. They’ll research the best gear they can afford and seek the highest quality music formats to play on it. That last bit has led to the c

thenextweb.com

PS - I can't locate it now, but I saw one test result where people *preferred* the 320k MP3 over lossless
PPS - here's one where they go for it by a small margin - https://www.reddit.com/r/audiophile/comments/620bxq

 

[freemansteve]

I am used to hearing live cymbals regularly and my son also played drums - what I do know is how differently crash cymbals sound from each other - some are inherently mushy and some are clear. It depends on the tastes of the drummer and how they are played, I think. Cymbals can sort of saturate your ears when played with verve so even live sound is not reliably good! I can easily hear all of that, but no matter how I try, I don't really hear a difference between cymbals MP3 and FLAC, at least on the (usually) quite 'busy' music I play most. Maybe it wouldn't bother me if I could!

It must be an old-age thing; at least I'll get a bus pass next year....

 

[antcollinet]

Anyone who wants to try a blind ABX between various bit rate MP3 and lossless, can do so here. I tried the 128kb/s test and was just guessing.

ABX Lossless vs Lame MP3 test list

A set of browser-based ABX tests to compare Lame MP3 encoded and lossless music samples

abx.digitalfeed.net

 

[solderdude]

I read somewhere the 'jangling keys' MP3 and WAV could be reliably told apart. No idea what sample rate and quality settings were used.
So, perhaps, some young folks with 20/20 hearing with specific music could maybe tell them apart in specific circumstances.

I'm old enough not having to worry about content above 15kHz. Maybe on higher compression rates some typical MP3 encoding artifacts could be heard by some (that know what to 'look' for).