A challenge for those who want one

[ferrellms]

I have read about audio file compression encoding and have always embraced it. After all, digital storage space and network bandwidth does have some (small) cost and why waste it if there is no difference in audio quality. The Fraunhofer people claim to have perfected MP3 encoding over the years, and have always used critical listening tests with experienced users to compare music (specially made test material is a different story) audio quality between CD quality and file compressed (MP3), and have found that higher bit rate MP3 files are audibly indistinguishable from the CD files they are encoded from despite the fact they are considerably smaller.

So I took the time to take one of my few remaining CDs (a modern one with good sound), ripped a tune from it to 4 different copies on the PC - CD quality WAV, 192kbs MP3, 320kbs MP3, and FLAC. I then sliced them up and reassembled the cut (using a common digital sound editor) into a copy of the original song in WAV format. The result was a single WAV file consisting of parts of 1) the WAV ripped from the CD 2) that same ripped from CD WAV file encoded into MP3 192 kbs and back into WAV 3) that same ripped from CD WAV file encoded into 320kbs and back into WAV and 4) that same ripped from CD WAV encoded into FLAC and back again - all spliced together.

I could not distinguish the transitions on high-end headphones and speakers. As far as I could hear, there was no audible difference in the different parts of the song that had been processed differently.

And yet, many claim to be dissatisfied with MP3 sound, calling it "compressed" and "lossy". Some misguided folks have claimed to prove there are problems with MP3 by directly comparing digital files and showing that the MP3 has less content in the audible range.

The issue is a conflation of file compression and audible compression. Of course MP3 files have less content - they are compressed to be smaller! But they don't SOUND any different. The encoding file compression/decoding playback processes utilize quirks in our hearing system to remove only content that cannot be heard in the context of the overall sound; ie, that is "masked" or "drowned out" by other simultaneous content that is louder.

"Lossless" has meaning the context of archival if not sound quality- a FLAC file can be expanded back into the original WAV without losing any bits whereas MP3 cannot.

The mass audiophile delusion of the audibility of modern music file compression is encouraged by the industry as there is money to be made. HD anyone?

Before you reply to this, try my experiment yourself! And please - you will just have to take my word for the quality of the original recording and my listening gear - I don't want to have to justify my judgments on those grounds. Try the experiment! If you can hear differences - good for you, perhaps you have exceptional hearing - go ahead and spend that extra money!

File sizes - WAV 80 MB, FLAC 50 MB, 320 MP3 18 MB, 192 MP3 11MB

 

[AudioStudies]

Although I can't prove it, and haven't run any tests, I disagree with some of your conclusions -- thus my opinion on things I have read in recent years. MP3 is not just "compressed" but compressed in a lossy manner. This is quite different from FLAC files which are compressed without loss. Its not just a matter of people calling MP3 lossy, it really is lossy (doesn't contain all the info). While I realize some people can't hear the difference, I don't think that I am one of them. I enjoyed myself a lot more when I upgraded my Tidal subscription beyond the MP3 level and to the higher level where I playback in 16 Bit that is not lossy. I don't think the analogy with High Res versus 16 Bit is valid, because both of those are not lossy. I agree with you that the High Res is not necessary and is marketing to increase sales.

 

[Frgirard]

The mp3 whose algorithm is one of the finest audio achievements of the previous decade.
I remember when HD track sold Beck music as hires and it was a mp3.
How much had heard the difference?

One major issue with the comparison is to locate a difference and to abx them.
Identify the difference is a work.

 

[AudioStudies]

The mp3 whose algorithm is one of the finest audio achievements of the previous decade.
I remember when HD track sold Beck music as hires and it was a mp3.
How much had heard the difference?

One major issue with the comparison is to locate a difference and to abx them.
Identify the difference is a work.

MP3 can still be considered a great achievement for what it is, but not for what it isn't. I agree some people will not tell the difference in ABX testing, but I think others will. I am "too retired" to do any testing, lol . . . you are right, it is "work" . . .

 

[RayDunzl]

So I took the time to take one of my few remaining CDs (a modern one with good sound), ripped a tune from it to 4 different copies on the PC - CD quality WAV, 192kbs MP3, 320kbs MP3, and FLAC.

I suspect the complaints with MP3 originated with lower data rates. 20 or more years ago.

Start with the lowest rate you can use to convert and work your way up from there.

 

[digitalfrost]

Could've save some work by doing an ABX test.

 

[Frgirard]

Could've save some work by doing an ABX test.

Time and sorrow but with the risk of a big narcissistic wound

 

[Blumlein 88]

Well complaints were mostly about low rate MP3. For awhile 128 kpbs was something of a benchmark. In formal testing 128 kbps is audible with good test tracks. AAC is said to be better than MP3, but I believe it was better at 128 kbps, and at higher rates there isn't much difference.

So in your test with worst being 192 kbps, that is getting to where many have trouble hearing a difference. Assuming your hearing is healthy (no hearing damage, you aren't old enough to have lost the treble etc.) you could probably hear a difference with some tracks at 192 kbps and almost surely could at 128 kbps.

Was listening to a stream from Accuradio yesterday and many tracks sounded pretty poor I thought. Well they are broadcasting in HE-AACv2 at 24-32kbps for its audio streaming. The same format mainly used for streaming video has been tested at 64 kbps as being close to, but not as good as 128 kbps AAC. No surprise such a low rate didn't sound good to me.

 

[ferrellms]

Although I can't prove it, and haven't run any tests, I disagree with some of your conclusions -- thus my opinion on things I have read in recent years. MP3 is not just "compressed" but compressed in a lossy manner. This is quite different from FLAC files which are compressed without loss. Its not just a matter of people calling MP3 lossy, it really is lossy (doesn't contain all the info). While I realize some people can't hear the difference, I don't think that I am one of them. I enjoyed myself a lot more when I upgraded my Tidal subscription beyond the MP3 level and to the higher level where I playback in 16 Bit that is not lossy. I don't think the analogy with High Res versus 16 Bit is valid, because both of those are not lossy. I agree with you that the High Res is not necessary and is marketing to increase sales.

Once again, "lossy" means different things for file and audio compression. As I said, you cannot reconstruct all the bits of a WAV file from an MP3 and you can from FLAC. So if you have some particular interest in retrieving all the bits from the original, use FLAC. If you just want something that is audibly indistinguishable from the WAV, and to save digital storage space or network bandwidth, use MP3 (320 kbs seems to be widely used these days).

Use your own ears.

 

[ferrellms]

Well complaints were mostly about low rate MP3. For awhile 128 kpbs was something of a benchmark. In formal testing 128 kbps is audible with good test tracks. AAC is said to be better than MP3, but I believe it was better at 128 kbps, and at higher rates there isn't much difference.

So in your test with worst being 192 kbps, that is getting to where many have trouble hearing a difference. Assuming your hearing is healthy (no hearing damage, you aren't old enough to have lost the treble etc.) you could probably hear a difference with some tracks at 192 kbps and almost surely could at 128 kbps.

Was listening to a stream from Accuradio yesterday and many tracks sounded pretty poor I thought. Well they are broadcasting in HE-AACv2 at 24-32kbps for its audio streaming. The same format mainly used for streaming video has been tested at 64 kbps as being close to, but not as good as 128 kbps AAC. No surprise such a low rate didn't sound good to me.

Well, professional audio journalists and Fraunhofer audio engineers with young ears could not hear any difference at 192 with music. You are right about 128 being the standard for a while and having some slight audible effects. I have heard differences with 192 and special test sounds. Most streaming is at 192 or 320 these days, I believe.

 

[ferrellms]

Although I can't prove it, and haven't run any tests, I disagree with some of your conclusions -- thus my opinion on things I have read in recent years. MP3 is not just "compressed" but compressed in a lossy manner. This is quite different from FLAC files which are compressed without loss. Its not just a matter of people calling MP3 lossy, it really is lossy (doesn't contain all the info). While I realize some people can't hear the difference, I don't think that I am one of them. I enjoyed myself a lot more when I upgraded my Tidal subscription beyond the MP3 level and to the higher level where I playback in 16 Bit that is not lossy. I don't think the analogy with High Res versus 16 Bit is valid, because both of those are not lossy. I agree with you that the High Res is not necessary and is marketing to increase sales.

You haven't run any tests... And yet you claim to hear a difference. Take the challenge, you will be surprised. I have been fooled often enough by expectation bias (everyone has it - it is part of being human) myself. I could swear I heard an improvement in my system when corrected to linear phase (all the sine waves that construct a sound are timed together at the ear, correcting for speaker driver spacing, etc.) So, I A/B tested it (I need a life obviously) and could hear no difference. Maybe I am partially deaf or whatever, but you need to actually hear it yourself in a direct, simultaneous comparison. It can be a humbling experience. But can save you money. Hate to say it, but Tidal is the prime example of making money from mythology and frailties in human perception. Google "expectation bias".

 

[q3cpma]

1) MP3 isn't that good, LAME is. Same with AVC and x264, the implementation usually matters a lot in these complex codecs.
2) MP3 is obsolete, AAC and Opus cream it. aotuv/Vorbis and Musepack consistently outperform it.
3) Most of these codecs are perceptually lossless at high enough bitrate without considering killer samples (e.g. applause), but most importantly, these are very rarely annoying when listening without looking out for defects.

 

[AudioStudies]

You haven't run any tests... And yet you claim to hear a difference. Take the challenge, you will be surprised. I have been fooled often enough by expectation bias (everyone has it - it is part of being human) myself. I could swear I heard an improvement in my system when corrected to linear phase (all the sine waves that construct a sound are timed together at the ear, correcting for speaker driver spacing, etc.) So, I A/B tested it (I need a life obviously) and could hear no difference. Maybe I am partially deaf or whatever, but you need to actually hear it yourself in a direct, simultaneous comparison. It can be a humbling experience. But can save you money. Hate to say it, but Tidal is the prime example of making money from mythology and frailties in human perception. Google "expectation bias".

I can't prove that it is not "Expectation bias" or a placebo effect, without running tests, but I can't possibly test every piece of music I want to listen to. And it just leaves me a bit unsettled that something is missing -- albeit you could be right that it is something I may not be able to hear $20 per month for Tidal is not exorbitant. As I build up my local library though, I may just use Plex without Tidal.

 

[RayDunzl]

Lest we forget:

MP3 hit public usage in 1993, maybe.

Your connection to the internet then maxed at 56 kilobits / second, maybe, I forget when 56k modems came about. I never had one. I did have broadband at work, since "we" were part of its development and deployment.

If you wanted to listen to a tune "real time" rather than wait for it to download, well, do the math.

3 minute tune at 320kb/s would take around 17 minutes to download.

Choose your poison.

 

[Blumlein 88]

Lest we forget:

MP3 hit public usage in 1993, maybe.

Your connection to the internet then maxed at 56 kilobits / second, maybe, I forget when 56k modems came about. I never had one. I did have broadband at work, since "we" were part of its development and deployment.

If you wanted to listen to a tune "real time" rather than wait for it to download, well, do the math.

3 minute tune at 320kb/s would take around 17 minutes to download.

Choose your poison.

Good point. Even with 56 k my phone line rarely let you connect at more than 28.8 k.

 

[abhijitnath]

https://www.npr.org/sections/therecord/2015/06/02/411473508/how-well-can-you-hear-audio-quality. You can just use this?

 

[danadam]

Assuming your hearing is healthy (no hearing damage, you aren't old enough to have lost the treble etc.) you could probably hear a difference with some tracks at 192 kbps and almost surely could at 128 kbps.

Actually some damage could help with that. I think I remember reading about someone who could hear mp3 artifacts because his hearing, due to his deficiency, was not following the model that the encoder was using.
Might have been an urban legend too

 

[AudioStudies]

Time and sorrow but with the risk of a big narcissistic wound

You should write song lyrics . . .

 

[Zensō]

My experience aligns with your conclusions. Hi-res, and now MQA, take advantage of audiophile obsessiveness to once again sell you the music you already own. In normal listening, 256 AAC and 320 Ogg Vorbis are audibly transparent to higher resolutions. It’s the idea of the “purity” of lossless that people are buying, not the actual sound quality.

 

[AudioStudies]

My experience aligns with your conclusions. Hi-res, and now MQA, take advantage of audiophile obsessiveness to once again sell you the music you already own. In normal listening, 256 AAC and 320 Ogg Vorbis are audibly transparent to higher resolutions. It’s the idea of the “purity” of lossless that people are buying, not the actual sound quality.

I never got on board with High Res or MQA, and don't think I need either one.