Is It Time To Rethink Lossless?

25 minutes ago, firedog said:

At least a label like NativeDSD tells you what the recording format was, so you can judge for yourself. When I buy from them, I ususally buy whatever the recording format was, and not a "conversion". (Even though I won't claim I hear differences between them). 

I love this transparency from NativeDSD. even if it doesn't tell us much in the big picture, it's honest and straight forward.

31 minutes ago, firedog said:

Led Zep? I'm assuming the 24/192 is a conversion from the master tape.

there is no indication from any other source that a 24/192 conversion was ever  done. It is 24/96 everywhere else. If there is a 24/192 master it seems highly unlikely Amazon is the only place that ever got it.

9 hours ago, Audiophile Neuroscience said:

I have just always thought (assumed) lossless relates to a codec, compression and decompression without loss of any information, whether that be perceived as important or not.

23 minutes ago, danadam said:

I agree.

23 minutes ago, danadam said:

Following from the previous, I'd say that "lossless"/"lossy" is just not applicable for WAV 🙂

It's a little stickier in my opinion because lossy MQA was packed in a FLAC container. Lossy compared to the other versions released for download and other streaming services, but in the big picture, all of them could be lossy compared to the master. 

I think the focus should be shifted more toward people and sound quality and away from classifications and numbers. If Bill Schnee recorded it and Doug Sax mastered it, the chances are very high it will sound great in all but the smallest bit rate MP3 that nobody releases anyway. Then there are artists and producers who prefer to squash dynamic range, which will sound good to them, but terrible to many at any sample rate, bit depth, bit rate, lossless container, etc...

Lossless should be used in terms of file conversions (compression or not). Using it in other places is outside of its definition, in my view. 

36 minutes ago, danadam said:

I thought that was the premise of the question, can you get back the 16/44 original from the 24/192 upsample. The assertion was (and apparently still is) that it's never possible, I showed that it sometimes is. I can agree that steep anti-imaging filter (99% bandwidth, the "-s" option) is probably not something that would be usually used, but I'm not sure if that's what you meant by calling it simple.

I certainly hear you. I think we should also consider the bigger picture as well, not just a lab type environment, even if the original premise was it can "never" happen. 

46 minutes ago, danadam said:

Could you explain what makes you think that I only compared peak and rms values of the 2 files? Because once again, that operation did compare each and every sample. Maybe if I split it into separate commands it will be more visible

the simple answer is you only showed peak and rms. Beyond that , I have no idea what it is you are doing. What program are you using, what are the settings? A compressed Flac from a Wav file without a conversion should result in an identical file, but again, what you posted does not indicate what you did or how you did it. At least it doesn't indicate it to me because I don't know how you are creating these reports.

IME the importance of lossless is directly tied to data density. Simple recordings with just a few instruments, MP3 is fine. Play a complex orchestral piece with massed instruments,

I crave DSD to keep instruments clean and separated.

4 hours ago, bbosler said:

the simple answer is you only showed peak and rms. Beyond that , I have no idea what it is you are doing. What program are you using, what are the settings? A compressed Flac from a Wav file without a conversion should result in an identical file, but again, what you posted does not indicate what you did or how you did it. At least it doesn't indicate it to me because I don't know how you are creating these reports.

He showed you exactly what programs and settings he used and the resulting output.

Perhaps a perusal of the applicable man pages might help.

Here's one to get you started:

https://linux.die.net/man/1/sox

34 minutes ago, kumakuma said:

He showed you exactly what programs and settings he used and the resulting output.

 

Perhaps a perusal of the applicable man pages might help.

Not "exactly."..... you are assuming one knew what sox is, Now that I do, I will check it out

IME, the worst offender for "being lossy" is the playback system. Including mine . The amount of detail that's effectively discarded is rather staggering, and this does most of the damage. Subjectively, that is. Has got better, a lot better, than the situation a couple of decades ago, but still has a way to go - to make it part of the mainstream. Every time you experience a rig that blows you away with the sense of place, immerses you totally in the action, and "shows you what's possible", you are being made aware of the degree of loss of most hifi playback.

So, first clean up your own backyard. Then start counting the angels on that pinhead ... .

6 hours ago, The Computer Audiophile said:

16 hours ago, Audiophile Neuroscience said:

I have just always thought (assumed) lossless relates to a codec, compression and decompression without loss of any information, whether that be perceived as important or not.

 

7 hours ago, danadam said:

I agree.

 

 

7 hours ago, danadam said:

Following from the previous, I'd say that "lossless"/"lossy" is just not applicable for WAV 🙂

 

 

It's a little stickier in my opinion because lossy MQA was packed in a FLAC container. Lossy compared to the other versions released for download and other streaming services, but in the big picture, all of them could be lossy compared to the master. 

This doesn't (conceptually) bother me. If I know the container is lossless with respect to the codec used I know the contents have not been subjected to losses. I do see your point that for many this may well be misleading if the contents start out as lossy.

6 hours ago, The Computer Audiophile said:

I think the focus should be shifted more toward people and sound quality and away from classifications and numbers. If Bill Schnee recorded it and Doug Sax mastered it, the chances are very high it will sound great in all but the smallest bit rate MP3 that nobody releases anyway. Then there are artists and producers who prefer to squash dynamic range, which will sound good to them, but terrible to many at any sample rate, bit depth, bit rate, lossless container, etc...

Chris, I think you probably know where I mostly stand on numbers based definitions of sound quality but its merely a matter of applicability and interpretation (another discussion).

The problem here as I see it is potential conflation of different concepts like impact of codec vs resolution and recording techniques and provenance etc.

"Lossless" may apply in the general sense of the word to each and every step of the process if talking sound quality - as in, this version has "lost" something in sound quality. The word lossy then loses all specificity other than an adjective for subjective perceived sound quality attributable to whatever (claimed) cause.

6 hours ago, The Computer Audiophile said:

Lossless should be used in terms of file conversions (compression or not). Using it in other places is outside of its definition, in my view. 

Lossless in terms of file conversion - if means "bit perfect after file conversion" . I think you're right it seems like a better use of the term lossless, and not a "focus shifted more toward people and sound quality". I see them as two distinct  issues.

This thread about defining lossless reminded me of another of Chris's comments re various Immersive audio technologies.

@The Computer Audiophile "This content is lossy, as discussed above. However, for 99% of the content, the lossy version is the highest version ever released to the public. I’d say we shouldn’t complain about a nonexistent product (lossless album ABC, XYZ, etc…) because it doesn’t exist outside the studio, but I’m sure there’s plenty of room for complaints."

Depending how lossless id defined, it would for example affect my preference for one or other immersive technologies.

Well, its like what kind of infinity we are talking about, adding a space like cd, dsd or the like to lossless would be great(not sure if I type it right but you get the point), also if it's the best quality available or original quality from recording. 

Meanwhile there are many misinformations about lossless for tws bluetooth buds, mentioning lossless-like cd quality like lossless cd quality. 

This is also a different topic but in camera field, there is a difference between lossless uncompressed and lossless compressed, which are not very lookalike - and I can see this thing also in audio, mostly when listening to compressed or not, devices can output audio cleaner when there is less computing for uncompressing. 

So my fav be it in audio or stills is uncompressed lossless. 

To up the game a bit I would say that hearing live is still better than most lossless, since there is not enough binaural recordings at original quality and that's how we hear without any loss. It's also a bit different from 360 audio, which adds a direction when our head moves. 360 could be added after the recording, of course it depends, bit I wouldn't be so sure about binaural.

There may be also trouble with amount of detail in each "lossless" file, as there exist upscaling too. 

I like end user upscaling tho, listening to wavy bass on my tws buds it makes music more enjoyable. 

Good listening.

I just remembered that Apple is also touting lossless audio for AirPods Pro with the new Vision Pro. But, the sample rate is 20 bit / 48 kHz. Not sure I’ve seen a 20/48 source yet :~)

If we were to redefine lossless as true to the absolute original source, we're gonna end up with very few actual lossless files...

2 hours ago, AudioDoctor said:

If we were to redefine lossless as true to the absolute original source, we're gonna end up with very few actual lossless files...

... and it is hard to believe a tape will be lossless 500 years after a recording was done.

I don't know about analog recording technique which would allow 100% perfect analog copies to retain the quality long term and thus not to depend on physical media used to store analog audio information. The only way to retain original analog recording quality as much as possible seems to be to use a digital format. With all the sampling nuisances (mirror images/aliases, ringing etc.). Miska would say that high rate DSD fits best for the purpose among the currently existing digital formats.

So true Chris.

To add to the confusion, some albums were conceptually created to be multichannel.
I'll never forget back when I first heard DSOTM in quad from DVD-A.
I realised that I had never really heard the album as it was intended!

So how can any album ever be lossless in stereo, if it was intended as a multi channel experience?

The term "lossless" is like the term "natural:" More about marketing and approach than a true objective meaning.

Windows Media Player used to use the term "Mathematically Lossless" in the mid 2000s when configuring the bitrate to rip CDs at. Clear and unambiguous.

I would like to point out something I suggested in https://andrewrondeau.com/blog/archive/2016-07

Quote

Thus, the final conclusion is that the term "lossless," (and "cd quality,") must be carefully defined, otherwise, it will be abused as a marketing term. "Lossless" can't become like the term "natural," "farm to fork," or "wholesome," where the retailer can redefine the description to match the product. The way to avoid confusion is to instead require terms like:

 

• 16-bit, Mathematically lossless: Basically, Flac from a CD
• Lossless at 120db: Completely lossless with a snr of 120db. The signal might be slightly different than mathematically lossless, though.
• Lossless at 96db: Completely lossless with a snr of 96db. The signal might be slightly different than mathematically lossless, though.

The big issue is that we don't objectively describe how lossy (and "lossless") audio compression algorithms manipulate audio. I tried to investigate it in the above-linked article.

An interesting observation, in hindsight, from my investigation is: The typical "lossy" codecs are still more accurate than an old-fashioned analog cassette tape. (And those Maxell XLIIs that I mixed in the 1990s with Dolby B sounded really good in my Mom's Ford Escort!)

We really need to rely more on objectively measuring audio codecs like I did in the above-linked article. These tests would compare the input and output, and report well-known metrics, like frequency response and signal-to-noise ratio. More importantly, we should be able to agree on a definition of lossless that an audio codec can meet, when objectively measured.

Lossless is lossless.  Full Stop.  Period.  The source doesn't matter.  Physical medium or a bunch of 1's and 0's on your computer or in the cloud or wherever.  When you encode an .MP3 into a FLAC (or other lossless format), the resulting file will be the exact same as the original.  You won't have lost anything more than the .MP3 gave you, to begin with.  Same as with a CD or any other medium.  All those "new numbers" that digital formats now have for Blu-Ray or whatever are just the resolution of the recording.

To me, there's nothing to rethink...

18 minutes ago, Luposian said:

Lossless is lossless.  Full Stop.  Period.  The source doesn't matter.  Physical medium or a bunch of 1's and 0's on your computer or in the cloud or wherever.  When you encode an .MP3 into a FLAC (or other lossless format), the resulting file will be the exact same as the original.  You won't have lost anything more than the .MP3 gave you, to begin with.  Same as with a CD or any other medium.  All those "new numbers" that digital formats now have for Blu-Ray or whatever are just the resolution of the recording.

 

To me, there's nothing to rethink...

That’s a very simplistic approach and certainly one that makes thinking about it easy. 
 

However, in the real world we have to use real examples and compare the source to the end product. The term lossless was simple when we just ripped CDs. Now it’s used everywhere to describe something. 
 

Do you think the following should be described as lossless?

1. MQA in a FLAC container.

2. CD quality 16/44.1 streaming files in ALAC or FLAC where the master isn’t 16/44.1. 

3. A 24/96 WAV version of an album that has a 24/352.8 WAV master. 

 

I actually never cared for the term lossless. You could take a 320kb file and copy it as “lossless” right? Which adds just enough compression to be able to decode back to 320kb.

this is all about where you draw the line of what is considered original source. So I would view lossless just as to what you as the copier is doing. And calling the source what ever source you are given before making the copy.

so again lossless to me is more the process or a verb, less so to describe a file. But I always went for aiff and wav as the files I wanted in my library.

I would  love to see an article about the best file formats today. Is FLAC still the gold standard for compatibility? I’m tired of my aiff and alac files not playing gapless on some devices like my car and oppo players. I’m considering converting to FLAC or WAV just not sure which is best. But I don’t like the term lossless so I tend to lean towards WAV which doesn’t use compression at all right?

you could also talk about “lossy” which the the bad word. But today, a cd is lossy if there is a higher resolution file available. Slippery slope if you ask me on where you draw these lines. There is enough to worry about in this hobby as it is.

28 minutes ago, Luposian said:

When you encode an .MP3 into a FLAC (or other lossless format), the resulting file will be the exact same as the original.  You won't have lost anything more than the .MP3 gave you, to begin with.

Looking further at this statement, it seems that you are totally fine if Spotify encodes its entire OGG Vorbis catalog into FLAC and calls it lossless. Is that what you said?

3 minutes ago, Deyorew said:

I actually never cared for the term lossless. You could take a 320kb file and copy it as “lossless” right? Which adds just enough compression to be able to decode back to 320kb.

 

this is all about where you draw the line of what is considered original source. So I would view lossless just as to what you as the copier is doing. And calling the source what ever source you are given before making the copy.

 

so again lossless to me is more the process or a verb, less so to describe a file. But I always went for aiff and wav as the files I wanted in my library.

 

I would  love to see an article about the best file formats today. Is FLAC still the gold standard for compatibility? I’m tired of my aiff and alac files not playing gapless on some devices like my car and oppo players. I’m considering converting to FLAC or WAV just not sure which is best. But I don’t like the term lossless so I tend to lean towards WAV

I like your thinking on this. 
 

Too bad all the streaming services advertise music as lossless. It really doesn’t tell the consumer much. 

51 minutes ago, The Computer Audiophile said:

Looking further at this statement, it seems that you are totally fine if Spotify encodes its entire OGG Vorbis catalog into FLAC and calls it lossless. Is that what you said?

Lossless, as I have always understood it, is "the source and destination are identical" (no audio information was truncated/removed, to make the file smaller, regardless of compression used or not).  Therefore, what you hear in the resulting file should sound identical to the source.

My hearing is pretty bad (I hear almost nothing in my right ear), yet I will always choose a lossless format over a lossy format.  Why?  Because I want to know that everything I'm hearing is what I can hear, not what someone else thinks I won't hear/miss audibly.  I'm a purist that way, even if I can't appreciate it.

The FLAC file, in your example, is a lossless version of the .OGG file.  Nothing MORE was lost in translation.  But you can't say that it's a lossless file, because  audio information was removed in the source file.  So, even though you used a lossless format, that doesn't make it a lossless result.  The only way you can consider the resulting FLAC truly lossless is if the source recording was uncompressed and no audio information was removed.

CD (source) -> WAV -> FLAC (end result) = truly lossless

CD (source) -> .MP3 -> WAV -> FLAC (end result) = not lossless (even though two of those file formats ARE lossless)

KNOWING there is audio loss and HEARING audio loss are two separate things.  I may never HEAR a difference between an .MP3 of a CD track and a FLAC of an audio track, but KNOWING something was removed, to make the .MP3, will always bother me.