Is lossy outdated in 2019 & onwards?

[q3cpma]

I've noticed they get unreasonably hostile when ever lossy isn't transparent on set of samples. Not the best place since they assume 100% lossless on a DAP is audiophile woo and assume that there aren't any artifacts in the bass/mids region.

Like who cares about Lame or others when 400 and 512GB cards are now cheap?. lol

512GB cards aren't that cheap, it's 85€ for a Samsung and 125 for a faster Sandisk, here. No other reputable brand available. And as those are always super slow, I don't wanna find how much time it'd take to actually copy 512 GB of data onto it in one go.

Hydrogenaud.io may have this annoying attitude, but they're right that avoiding lossy (whatever the bitrate (!)) purely for quality reasons is mostly for audiophiles who never did ABX these "obvious artifacts" and all the bullshit you can usually hear. Using a good codec with a very adaptative VBR implementation (Vorbis or Opus, for example) helps those problematic samples/genres a lot.
I'd never use lossy I didn't encode personally, on the other hand.

 

[BDWoody]

Those of us who steam music over the internet.

Well THERE'S yer problem...

 

[Julf]

Well THERE'S yer problem...

How is that a problem?

 

[BDWoody]

How is that a problem?

Joke...implying literal steam rather than the intended stream.
Sorry...pretty obscure...

 

[Julf]

Joke...implying literal steam rather than the intended stream.
Sorry...pretty obscure...

Ah, didn't notice the typo! Thanks!

 

[yikky900]

512GB cards aren't that cheap, it's 85€ for a Samsung and 125 for a faster Sandisk, here. No other reputable brand available. And as those are always super slow, I don't wanna find how much time it'd take to actually copy 512 GB of data onto it in one go.

Hydrogenaud.io may have this annoying attitude, but they're right that avoiding lossy (whatever the bitrate (!)) purely for quality reasons is mostly for audiophiles who never did ABX these "obvious artifacts" and all the bullshit you can usually hear. Using a good codec with a very adaptative VBR implementation (Vorbis or Opus, for example) helps those problematic samples/genres a lot.
I'd never use lossy I didn't encode personally, on the other hand.

So the british company integral who came out with the first 512GB is not worthy?.

Lossy has hard limits there no guarantee the 320 to 512kb/s settings will crush a killer sample. There a few threads on HA that show that and why they pick lossless in that case.

Those of us who steam music over the internet.

So what?, What that gotta do with offline users who use their phone/DAP. Cute that you ignore Amazon HD/Tidal are options in that case. No clue why you get offended that some want full quality on there source while reminding Vinyl users why there format is flawed with no irony.

 

[Julf]

So what?, What that gotta do with offline users who use their phone/DAP. Cute that you ignore Amazon HD/Tidal are options in that case. No clue why you get offended that some want full quality on there source while reminding Vinyl users why there format is flawed with no irony.

I was simply answering the (rhetoric) question in "Like who cares about Lame or others when 400 and 512GB cards are now cheap?. lol". Not sure I was the one who got offended...

 

[Sal1950]

Can't believe you guys are still arguing over this.
Yes, lossy codecs are obsolete.
Unless you need them, then they are valuable.

 

[q3cpma]

So the british company integral who came out with the first 512GB is not worthy?.

Yep, only buy from actual makers, not glorified intermediaries. Are you sure about your claim, by the way? You'd expect actual OEMs like Samsung, Sandisk or Toshiba to make them first.

Lossy has hard limits there no guarantee the 320 to 512kb/s settings will crush a killer sample. There a few threads on HA that show that and why they pick lossless in that case.

By killer sample, do you mean musical content? Because I'm quite skeptical about 320-512 kbps AAC/Vorbis/Opus/Musepack failing over anything that's not made to defeat them.

 

[daftcombo]

We know about frequency masking and threshold of audibility more than before. The question should rather be: is lossless outdated in 2019? imho.

 

[Blujackaal]

Isn't that the forum, That overhyped AAC to high heavens since they say QAAC is fine at 128kbps. But they're at least 25+ killer samples for it that break it at 320kbps, Yet Vorbis at Q9+ has no such issue?. Weird there now hyping xHE AAC <128Kbps despite people only caring about Vorbis(160+ kbps) & LAME(190+ kbps) if they go Lossy or use streaming stuff.

Both AAC/Opus stuffer from pixel like artifact like bad Sat signal, under <160k QAAC & <112kbps Opus.

 

[pfzar]

Lossy codecs will be alive and well for the rest of my lifetime. AC4 and MPEG-H are lossy codecs carrying around 30 to 50 audio tracks. They are being used in ways for audio-only consumption eg. Dolby Atmos and Sony 360. They are also of course used for OTT television and movies.

 

[Wes]

what we should do is encode things so they overlap in the bit stream - just as some bacteria and viruses use overlapping encoding of genes in their DNA or RNA.

 

[stevenswall]

lossy files that are (99% of the time) transparent because of the efforts of lots of people, and because their efforts were verified by the folks over at Hydrogenaudio.

It does cover almost all music without issues, that's why people don't care or defend it as being transparent is the vast majority of cases.

I'm open to being wrong though if the majority of people can pass tests comparing high quality lossy streaming codecs with lossless stuff:

http://abx.digitalfeed.net/list.html

 

[stevenswall]

what we should do is encode things so they overlap in the bit stream - just as some bacteria and viruses use overlapping encoding of genes in their DNA or RNA.

Like having a lossy copy over the top of a lossless one? It would be cool if there was an encoder that made ~100 different overlapping versions of a thing. Play all 100 and you're listening to the lossless file. Play 90 of them and it's super high quality streaming. Play half of them and it's like a regular 320kbps mp3 file, and then as you go down maybe have it only focus on what's the loudest.

One time I had a lossless encode and subtracted a lossy encode from it in Audacity. All that was left was extremely quiet and sounded like warbling water noises. Don't think I'd hear any of that if either version was playing.

 

[q3cpma]

what we should do is encode things so they overlap in the bit stream - just as some bacteria and viruses use overlapping encoding of genes in their DNA or RNA.

That's what the MDCT used by most modern codecs is about.

 

[q3cpma]

Like having a lossy copy over the top of a lossless one? It would be cool if there was an encoder that made ~100 different overlapping versions of a thing. Play all 100 and you're listening to the lossless file. Play 90 of them and it's super high quality streaming. Play half of them and it's like a regular 320kbps mp3 file, and then as you go down maybe have it only focus on what's the loudest.

One time I had a lossless encode and subtracted a lossy encode from it in Audacity. All that was left was extremely quiet and sounded like warbling water noises. Don't think I'd hear any of that if either version was playing.

The concept really isn't new. See https://en.wikipedia.org/wiki/Bitrate_peeling or just what JPEG2000 does.

 

[Wes]

Like having a lossy copy over the top of a lossless one? It would be cool if there was an encoder that made ~100 different overlapping versions of a thing. Play all 100 and you're listening to the lossless file. Play 90 of them and it's super high quality streaming. Play half of them and it's like a regular 320kbps mp3 file, and then as you go down maybe have it only focus on what's the loudest.

One time I had a lossless encode and subtracted a lossy encode from it in Audacity. All that was left was extremely quiet and sounded like warbling water noises. Don't think I'd hear any of that if either version was playing.

no this is more complicated - a codon starts at a different place on each xNA 'stream' - that allows the different codons to overlap, with zero add'l bits

 

[stevenswall]

The concept really isn't new. See https://en.wikipedia.org/wiki/Bitrate_peeling or just what JPEG2000 does.

The problem/point isn't with the concept being old or new. That does nothing for us.

Is there a streaming service or mainstream encoder that uses the concept? If it's eighteen octillion years old, it's still worth complaining about/talking about/demanding if companies aren't building something based on it. Glad they're using that for photos, though that seems like the worst application since nobody streams photos. Perhaps there's a video codec that does that though. I wish Netflix would use it and smoothly scale resolution instead of clicking back and forth between blurry and not.

 

[q3cpma]

The problem/point isn't with the concept being old or new. That does nothing for us.
Is there a streaming service or mainstream encoder that uses the concept?

No, and for good reasons. Current lossy codecs are just enough for 99% of the population, simple enough to propose FLAC at a price than implement a complex scheme just to please such a tiny minority.

nobody streams photos.

Huh? And what are all the pictures you see in your web browser?