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

[solderdude]

Didn't GoldenSound use square waves or something? For a normal song I should not be able to pass the test, in theory. Do you know where I can find the files?

There were also known music pieces on it.

 

[escape2]

I don't know what it is with the beginning of this song that the codecs seem to struggle with

Is it the same story with AAC?

 

[Tangband]

Stereophile has done measurements between 320 kbps and red book CD.
Its about 45 dB difference - advantage red book CD.
The dynamic range with 320 kbps is only 51 dB with those test tones.
This is really bad. The red graph is 320 kbps, green is CD.
Why would someone pay money for this music-destruction ?

Read more here:

MP3 vs AAC vs FLAC vs CD Page 2

For reference, fig.3 shows the spectrum of the signal on the CD. Other than the well-defined green vertical lines representing the tones and the uniform background noise, the spectrum is clean. Important points to note with this graph are that 1) all musical fundamentals lie to the left of the...

www.stereophile.com

 

[Julf]

Stereophile has done measurements between 320 kbps and red book CD.
Its about 45 dB difference - advantage red book CD.
The dynamic range with 320 kbps is only 51 dB with those test tones.
This is really bad. The red one is 320 kbps, green is CD.
View attachment 172858View attachment 172859
Read more here:

MP3 vs AAC vs FLAC vs CD Page 2

For reference, fig.3 shows the spectrum of the signal on the CD. Other than the well-defined green vertical lines representing the tones and the uniform background noise, the spectrum is clean. Important points to note with this graph are that 1) all musical fundamentals lie to the left of the...

www.stereophile.com

Do you think those measurements make sense with perceptual encodings? Anyway, where do you get the 51 dB from? The piece you linked to says:

But the background noise components, which on the CD all lay at around –132dB, have all risen to the –85dB level. With its limited bit budget, the codec can't encode the tones without reducing to almost half the 16 bits of CD resolution.

By the way, 85 dB is 14 bits, not half of 16.

 

[escape2]

Why would someone pay money for this music-destruction ?

Which paid streaming service uses 320 kbps MP3 format?

Also, this article is almost 14 years old. I'd be curious to know how much has changed since then. The Fraunhofer MP3 codec the article tested hasn't been developed since 2003, but there are other MP3 codecs widely used today that have undergone a lot of development during this time.

 

[Tangband]

By the way, 85 dB is 14 bits, not half of 16.

You need to read the graph properly. Max level is at -40 dB in the graphs when the test is done. This is maximal level.
The 320 kbps result is aprox only 8,5 bit resolution. Its not hifi !

 

[Julf]

You need to read the graph properly. Max level is at -40 dB in the graphs when the test is done. This is maximal level.
The 320 kbps result is aprox only 8,5 bit resolution. Its not hifi !

You can'yt really determine dynamic range from a graph like that.

 

[Tangband]

Which paid streaming service uses 320 kbps MP3 format?

Also, this article is almost 14 years old. I'd be curious to know how much has changed since then. The Fraunhofer MP3 codec the article tested hasn't been developed since 2003, but there are other MP3 codecs widely used today that have undergone a lot of development during this time.

You have a point, but why defending something thats clearly NOT good ? The lossless alternativ cost the same or almost the same.

 

[Julf]

You have a point, but why defending something thats clearly NOT good ? The lossless alternativ cost the same or almost the same.

Read back on this thread. No, especially if you are streaming, the cost is not the same.

 

[HRTF_Enthusiast]

Is it the same story with AAC?

I was able to pass the test with AAC as well, but none of them are as extreme as mp3 at that part. I'm not sure what is better between 320 kbps OGG and 256 kbps AAC.

 

[somebodyelse]

I'm not sure what is better between 320 kbps OGG and 256 kbps AAC.

Ogg is the wrapper format not the codec. I guess you're talking about Vorbis or Opus, but it can contain FLAC too, among others.

 

[HRTF_Enthusiast]

Ogg is the wrapper format not the codec. I guess you're talking about Vorbis or Opus, but it can contain FLAC too, among others.

Oh, well idk. I just exported the flac as the level 9 quality setting for OGG in audacity because the wiki said it's 320 kbps.

 

[krabapple]

Is ASR the place to rehash old conclusions again and again? It surely seems that way.

It's been known at design time (in the 1990s, when 5 min of redbook PCM would completely fill a regular HDD, and it would take hours to compress it to 128kbps mp3) that MP3 would have audible artifacts with music with short transients (like castanets or claps-they become smeared), high frequency content (higher than 16kHz are either removed or wrong).

LAME -V 0 has no lowpass filtering. What's it doing wrong up there?

 

[danadam]

Didn't GoldenSound use square waves or something? For a normal song I should not be able to pass the test, in theory. Do you know where I can find the files?

3.32 GB file on MEGA

mega.nz

(from https://audiophilestyle.com/forums/...al-to-test-mqa-mqa-deep-dive/#comment-1126044 )

 

[krabapple]

I think mp3's are the 30" CRT TVs. There have been many improvements like AAC, Vorbis, Opus, or whatever. Just like there have been 30"+ LCD TVs forever. But now, nobody buys 30" LCD TV's. You either stick with, and complain about the 30" Trinitron, or move on and get lossless (50" LCD).

What irks me the most is that people believe these are the only two choices. They are not!

I think a further issue for any still using lossy encoding, beyond just being aware that MP3 alternatives exist, is not knowing what hardware/apps support AAC Vorbis Opus Musepack.

'Everything' supports MP3. So they stick with what they know. And with the right settings , file size can still be so low and sound quality can still be so high it's a nonissue (most listeners will fail an ABX test at these settings unless they undergo training with carefully chosen samples), even if there are 'better' lossy codecs out there (higher quality at lower bitrates).

Arguably if a listener cares *that much* about 100% bulletproof transparency but still wants smaller files, they'll just use lossless compression, which is also widely supported now.

 

[dwkdnvr]

Stereophile has done measurements between 320 kbps and red book CD.
Its about 45 dB difference - advantage red book CD.
The dynamic range with 320 kbps is only 51 dB with those test tones.
This is really bad. The red graph is 320 kbps, green is CD.
Why would someone pay money for this music-destruction ?
View attachment 172858View attachment 172859
Read more here:

MP3 vs AAC vs FLAC vs CD Page 2

For reference, fig.3 shows the spectrum of the signal on the CD. Other than the well-defined green vertical lines representing the tones and the uniform background noise, the spectrum is clean. Important points to note with this graph are that 1) all musical fundamentals lie to the left of the...

www.stereophile.com

This is pretty much a completely bogus test though. Multi-tone sine waves may be useful for testing electronics, but not perceptual codecs - they don't really resemble music at all, and the algorithms are going to struggle since they rely on the perceptual models of music (i.e. harmonically related information) to do their work.
Furthermore, the 'noise floor' of redbook isn't at -132 - that's an artifact of the FFT lengths used in the analysis. 96dB is all you get.

It also misses (IMHO) the benefit of this discussion. Nobody is claiming that 320K lossy is audibly transparent. What is interesting is
- it's remarkably close for throwing out 75% of the raw data
- it probably IS audibly transparent on a fair bit of program material
- it's a perfect avenue to introduce yourself to ABX testing and listener training
- (potentially) for those that do use streaming services, is there a way to minimize the damage

IMHO we should be encouraging all participants on this site to engage in some ABX testing to gain some direct experience in some of the ideas that are the foundation of the 'science' since IMHO that gets a bit lost. IMHO the important 'science' on this site is NOT the measurements of the gear, but rather the results from psychology showing just how flawed and unreliable our perceptions are, and how bias (both conscious and unconscious) can so easily influence the subjective experience. ABX testing lossy compression where we 'know' there are differences can be something of a reset of expectations around the audiophile experience by showing just how close they are in many cases.

 

[krabapple]

Stereophile has done measurements between 320 kbps and red book CD.
Its about 45 dB difference - advantage red book CD.
The dynamic range with 320 kbps is only 51 dB with those test tones.
This is really bad. The red graph is 320 kbps, green is CD.
Why would someone pay money for this music-destruction ?
View attachment 172858View attachment 172859
Read more here:

MP3 vs AAC vs FLAC vs CD Page 2

For reference, fig.3 shows the spectrum of the signal on the CD. Other than the well-defined green vertical lines representing the tones and the uniform background noise, the spectrum is clean. Important points to note with this graph are that 1) all musical fundamentals lie to the left of the...

www.stereophile.com

Typically misleading garbage that Stereophile should have been ashamed to publish even in 2008.

Leaving aside the reliance on mostly 128kbps Fraunhofer, and the panty-twisting about noise at -85dB and lower using *pure tones* as inputs to make pronouncement on DR -- there's not a single controlled music/voice listening test in evidence, which is how lossy codecs *must* be evaluated, as that is the point of codecs based on psychoacoustic models.

In terms of sophistry, this is no different than George Massenberg back in the day playing the 'difference' signal of lossless vs lossy to aghast audiences to 'demonstrate' how 'bad' MP3 is.

 

[LeftCoastTim]

Ogg is the wrapper format not the codec. I guess you're talking about Vorbis or Opus, but it can contain FLAC too, among others.

sadly, vorbis encoder that output .ogg audio stream is called "oggenc2".

 

[LeftCoastTim]

You need to read the graph properly. Max level is at -40 dB in the graphs when the test is done. This is maximal level.
The 320 kbps result is aprox only 8,5 bit resolution. Its not hifi !

Yeah, perceptual codecs use the fact that sound is masked when other sounds exist. It really does ruffle a Golden Ear's feathers to show them their real hearing ability instead of their imagined ones.

 

[LeftCoastTim]

I don't know what it is with the beginning of this song that the codecs seem to struggle with
View attachment 172809
https://www.mediafire.com/folder/gi11cs4urjjd1/flac+vs+ogg
Here are the two files

You likely have exceptional hearing. How old are you by the way? In decades only (20s/30s, etc).

Also, if you are looking out for a particular "signature", like high hats sounding dull or crispness of attacks, or some other thing, then you know what you are sensitive to. And watch your ability degrade over the next decade as you can't abx them anymore. Been there, done that

And now that you can abx 320kbps vorbis, what's your *perceptual* score of it?
5. transparent, 4. perceptable not annoying, 3. slightly annoying, 2. annoying, 1. very annoying?

Here is another person with exceptional hearing who can abx vorbis -q9, but it was hard for them.