Where is the AI audio upscaler?

[litemotiv]

@litemotiv and after all of this time we still have it (to a less extent than slapping predator as cat but still there).
By the way that's the reason why so far all of them (attempts) failed (felt sibling and harsh in highs where they added the most).

Yes stress due to increased anomalies in ambient noise is also a leading cause of increased mental problems worldwide:

Noise Pollution Isn’t Just Annoying — It’s Bad for Your Health

Whether it’s the hum of an always-on TV or the window-rattling noise of airplanes overhead, the din of modern life poses a threat to mental and physical health.

www.brainfacts.org

 

[ZolaIII]

@litemotiv stress due to ambient and tempo of life, for which music is a great therapy.

Edit: lucky we are very adoptable basterds (abuse is abuse never the less).

 

[ThatM1key]

Yeah lets get back to the upscaling. Years ago mainly, if you want a very sharp looking image you had to do downscaling (if you had a powerful gpu). Downscaling was your next option after "max settings" to achieve that extra crisp image.

Like @litemotiv said above. Our brains are very good at filling in details. This image is a good example. If you couldn't tell, this is GTA 5 with special 'filters".

 

[Galliardist]

Yeah lets get back to the upscaling. Years ago mainly, if you want a very sharp looking image you had to do downscaling (if you had a powerful gpu). Downscaling was your next option after "max settings" to achieve that extra crisp image.

Like @litemotiv said above. Our brains are very good at filling in details. This image is a good example. If you couldn't tell, this is GTA 5 with special 'filters".
View attachment 181244

I suspect that our brains are somewhat more hesitant about filling in audible details than visual, though.

 

[Katji]

I suspect that our brains are somewhat more hesitant about filling in audible details than visual, though.

Audiophiles seem to be less hesitant about it.

 

[EdTice]

Not even close to a realistic comparison. Numerous well-controlled blind tests have shown that almost no one can hear the difference between high bitrate AAC and Vorbis compared to Redbook and higher, particularly if not isolating very small snippets or reverb trails. 1080P compared to 4K is easy for anyone to spot.

4k is easy for some people to spot. There are plenty of instances of people talking about how nice their 4K playback looks and nobody having the heart to tell them that they are looking at a 1080p picture.

 

[litemotiv]

4k is easy for some people to spot. There are plenty of instances of people talking about how nice their 4K playback looks and nobody having the heart to tell them that they are looking at a 1080p picture.

I guess it depends on the size of the TV and the viewing distance whether it's really noticeable?

 

[EdTice]

I guess it depends on the size of the TV and the viewing distance whether it's really noticeable?

My TV is 75" and I view from about 3.0m so I'm right on the cusp. I could probably differentiate if I looked carefully. If somebody secretly switched the resolution on me, I might not notice it. I am unfortunate to notice JPG/MPEG artifacts, though, since I worked with image processing early in my career. I wish I could remove that capability from my brain!

 

[Trell]

I guess it depends on the size of the TV and the viewing distance whether it's really noticeable?

Exactly, and this reminds of the 720p vs 1080p debates of yore. I'm sitting 2.7 meters from an 55" LG OLED CX.

HDR and extended color space is noticeable, and that is one of the major benefits of getting an UHD display.

 

[Zensō]

I guess it depends on the size of the TV and the viewing distance whether it's really noticeable?

Yeah, I can understand that with TVs. I'm not a big TV watcher, so I was thinking in the context of computer monitors, which are almost always viewed at close range.

 

[ZolaIII]

My TV is 75" and I view from about 3.0m so I'm right on the cusp. I could probably differentiate if I looked carefully. If somebody secretly switched the resolution on me, I might not notice it. I am unfortunate to notice JPG/MPEG artifacts, though, since I worked with image processing early in my career. I wish I could remove that capability from my brain!

Our brain tries to protect pore user and preserve and save energy.
Same thing GPU industry is perusing to mimic (over sampling or supper sampling is actually something else and useful in the same way as EQ-ing [example; extrapolation and extraction of gama, smoothing...]) but real purpose of DLSS and such is to mask what you won't see (or performance trade off what you barely could if you try really, really hard) and easy up the workload on GPU.
Artifacts pop up (as potential threth, unknown, anomaly) while it's the same picture disregarding of granulate small details (resolution) which are discarded first. Same thing goes for HDR and other types of shedding light mapping. And as always it's subject related (objectively and subjectively) same as audio.

 

[BoredErica]

I sit right here lol.

 

[EdTice]

View attachment 181835
I sit right here lol.

For your TV or computer monitor?! Your friends probably do not want to come over to watch the Superbowl!

 

[BoredErica]

For your TV or computer monitor?! Your friends probably do not want to come over to watch the Superbowl!

I'm not into sports, and I have no friends IRL who watch movies. :^)
Might be moving to a 42in OLED TV soon, but will be sitting at the same ratio/FOV away from the TV, so pretty close.

 

[ZolaIII]

I'm not into sports, and I have no friends IRL who watch movies. :^)
Might be moving to a 42in OLED TV soon, but will be sitting at the same ratio/FOV away from the TV, so pretty close.

Actually that's quite OK. Even this should only be a guideline.

TV Size To Distance Calculator (And The Science Behind It)

Choosing a new TV for your room can be a daunting challenge. The market has never been more complicated, with dozens of new models released each year and a mountain of marketing jargon to work through.

www.rtings.com

 

[litemotiv]

Actually that's quite OK. Even this should only be a guideline.

Yes those charts are based on optical resolution with 20/20 vision, there are indeed people with sharper eyesight such as 15/20, or in rare cases even better:

It seems that the best eyesight ever reported in a human was in an Aborigine man with 20/5 vision! To give you an idea of how clear and far he could see, his vision measurement compares to the natural sight of eagles. From 20 feet, he could perceive the fine details that most people can only see from 5 feet away.

I wonder how this translates to hearing. Apart from higher frequencies, do some people hear 'better' in the standard frequency range than others? How would this work?

 

[ZolaIII]

Yes those charts are based on optical resolution with 20/20 vision, there are indeed people with sharper eyesight such as 15/20, or in rare cases even better:

It seems that the best eyesight ever reported in a human was in an Aborigine man with 20/5 vision! To give you an idea of how clear and far he could see, his vision measurement compares to the natural sight of eagles. From 20 feet, he could perceive the fine details that most people can only see from 5 feet away.

I wonder how this translates to hearing. Apart from higher frequencies, do some people hear 'better' in the standard frequency range than others? How would this work?

As usual you should be more concerned with wors eyesight than Eagle eyes (same like with golden ears) as that will be far more the case.
There are people who hear considerably better and its not about how better they really can hear it but how much they process the information. For example blind people. Extensive training can help you to layer the certain frequency ranges better (vocals mostly) and disregard rest while doing so (or just do it purposely because you need it and and it's easier by rising that range for a level or two which you can perceive).

 

[EdTice]

I'm not into sports, and I have no friends IRL who watch movies. :^)
Might be moving to a 42in OLED TV soon, but will be sitting at the same ratio/FOV away from the TV, so pretty close.

In that case, you have plenty of time to measure some audio equipment!

 

[Sam Ash]

For video, TVs have them for a while now, Nvidia and AMD have it on their GPUs,, even a streaming stick (Nvidia Shield) is capable of real time AI upscaling with pretty impressive results.
So why doesn't anyone offer this for simple music files yet? I believe it would be a godsent for the bandwith issues and produce exceptional results for files that don't even have lossless never mind high res originals.

And I can't even think of why this would be hard to make. Just pick a few ten thousand music files with high res. Compress them with lossy formats. Feed the training data and expected results to a neural network and voila - there is the AI model with 99,999% success rate in upsampling lossy music to their original state. I think you could train this in a few days on a consumer GPU even.

I guess video upscaling is useful and definitely more readily noticeable when compared to audio. For example 1080p content being upscaled to 4K which is where the bulk of upscaling happens with the occasional lower resolution sources such as 720p. Some of the more sophisticated interpolation algos do an impressive job of it as it is combined with edge enhancement technology whilst keeping noise to an absolute minimum. Another aspect I appreciate, as a personal preference, is good frame interpolation. However, that may be a thing of the past as we step into an era of high frame rates which I think will add value to overall image fidelity. Apart from upscaling or when upscaling is not required (for example when watching 4k content on a 4K output device), I would still love to use refined edge-enhancing sans noise algos along with dynamic local enhancement algos which I think are being refined or looked into by developers of high-end video processors. It would be nice if Amir can review these at some point.

In terms of audio, I think many factors come into play to influence sound beyond just file resolution. Effective upscaling is definitely beneficial but what other enhancement technologies out there make an audible difference when listening to good sources on a good system? I agree, resolution and detail are important but the perception of audio resolution can be a misleading because a slight elevation in the higher frequencies can be perceived as better detail and higher resolution. This maybe pleasing to begin with but prolonged listening usually results in ear fatigue. Just like the eye, the ear is highly adaptive so that needs to be taken into consideration.

 

[Sam Ash]

I've noticed the fact that Audirvana is appreciated within the audiophile communities and music lovers. Any users here? It would be nice to have your thoughts in terms of the difference it makes. Amir, have you tried it out?