Slashdot Mirror

Error correction is actually done with extra bits stored on the cd (and actually about 5 bits per 16-bit sample, I think) and dithering actually *adds* dynamic range to about 120 dB at frequencies that matter (at the expense of a bit more noise at high frequencies above 10 kHz).

https://people.xiph.org/~xiphm...
https://youtu.be/cIQ9IXSUzuM

The compact disc has a theoretical max 95 dB dynamic range ... Add it all up and you get a usable DR of about 60 - 65 dB

You are way wrong

> "They're both equivalent". No, they aren't.

You're asking the wrong question. You should ask, "is there sufficient information to describe the analog signal?"

In the case of a band-limited signal, sampling above the Nyquist frequency at 16 bits or better is sufficient for audio...ANY audio. More information will not improve your results.

In fact, having more information can actually make it sound worse, not better.

Maybe relevant for you - https://people.xiph.org/~xiphm...

Thanks for the link, my own more limited experience jibes with that OK. My audio listening equipment comes down to Sennheiser HD420s plugged into an older M-Audio Mobile Pre USB, which is in turn plugged into the low-noise USB DAC plug on my PC, which plugs into a Tripp Lite isobar 4... I also have a Kenwood and some Yamaha monitors, but I don't have them hooked up ATM.

Maybe relevant for you - https://people.xiph.org/~xiphm...

Another Anonymous Coward noted:

because, well, why not? 10 times better tweeter sounds and a noise floor so quiet your ears will go DEAF and more channels that Amsterdam has hooker hangouts along the canals. But then there's this poophead here,

https://people.xiph.org/~xiphm...

I make it a rule not to waste mod points on AC comments. I broke that rule for this one, because the amount of technical detail in the link it included, and the links it contains to tools that will let interested parties actually test many of the assertions its author makes are, in fact EXTREMELY informative. Anyone who cares about audio quality for end-user playback needs to understand the scientifically-tested-and-confirmed facts about the physiology of human hearing and the effects of various sampling rates on user-detectable qualitative differences between them.

The article linked does a superb job of providing the factual information that's necessary to understand both the technical issues involved in audio sampling, mixing, mastering, and playback, and the experimental bias that colors most anecdotal observations about the physiological capacity of humans to distinguish between them during playback. If you want to discuss the subject with any real level of technical expertise, you really need to understand the issues it dissects ...

(Posting as AC only so as not to undo prior upmods in this thread.)

--

Check out my novel ...

because, well, why not? 10 times better tweeter sounds and a noise floor so quiet your ears will go DEAF and more channels that Amsterdam has hooker hangouts along the canals. But then there's this poophead here,

https://people.xiph.org/~xiphm...

Into the black with him!

Mozilla employs people from Xiph such as Chris Montgomery, Timothy Terriberry, Jean-Marc Valin, and Thomas Daede. I don't think paying the bills is laughable. Mozilla has funded development of Opus, Daala, and AV1.

If it helps, here's a recent blog post from Chris Montgomery on AV1's contstrainted directional enhancement filter.

Data plan on biggest Spanish carrier [finder.com] is 15 Euro for 1.5 GB. Or 1 Euro per 100 MB. That's probably about the size of the sound samples which would need to be transmitted back each month.

Let's do the math. State-of-the-art audio codecs, such as Opus, can intelligibly store speech using as little as 0.7 Kb/s. The perceived quality at such rates is terrible, but it may be good enough for the purpose of fuzzy matching to a known broadcast signal. And the device doesn't need to be recording all the time -- it needs to be recording only at or around the known broadcast times. So recording and transmitting a single two-hour game would require only 615 KB. A hundred megabytes is enough to transmit about 15 such games. I don't follow soccer so I have no idea whether or not that's a typical number of games for Spanish teams to play in a month.

These minute lost data is often what idiots attribute warmth or depth to the music.

Fixed that for you.

The quantization artifacts in 16-bit PCM audio are at approximately -96dBFS, in other words way below any audible level. If you do amplify an otherwise silent section up so the noise is audible, you'll find that it's more less just benign white noise, aka "tape hiss".

Analog recordings are also "approximations of the actual sound", except they're much much worse than digital recording, due to nonlinear frequency response, crosstalk, speed variation and a host of other issues.

It is blindingly obvious that you have no idea how digital sampling works. Educate yourself: https://xiph.org/video/vid2.sh...

You might want to have a discussion with Neil Young. He explains in great detail what's wrong with CD, and if you run it to ground he's right.

No, Neil is wrong. Monty from Xiph explains why, in great detail:

https://people.xiph.org/~xiphmont/demo/neil-young.html

You totally misunderstand how the DAC recreates the original signal. Watch this educational video linked above: https://xiph.org/video/vid2.sh...

First off, that really only gives you 2 samples per WAVEFORM at 20 kHz.

Watch the video. It demonstrates how 2 samples is enough.

Nice video, thanks! Didn't learn anything I didn't already know, except for the term "Gibbs Effect". I was familiar with the effect, just not the name.

HOWEVER...

The DAC doesn't "recreate the original signal". The DAC puts out Discrete STEPS (despite what the video claimed). That is ALL that a DAC does, period. They do not produce "Lollipop" output.

It is the Dithering (a/k/a digital noise) h/w and the "Reconstruction Filter" that is mostly responsible for attempting to smooth-out those STEPS, and remove aliasing and other artifacts.

So, I guess what I am really trying to point out is best demonstrated by the "Gibbs Effect" demonstration. Because music is very rarely all sine waves, it is the higher than 20 kHz harmonics that suffer from the 44.1 ks/Sec sample rate of CDs, and why cymbals sound like escaping steam, and tambourines make me want to scream on them, too.

IOW, I stand by my original statement that 44.1 ks/Sec is simply NOT enough, period; because we don't listen to sine waves, generally.

You totally misunderstand how the DAC recreates the original signal. Watch this educational video linked above: https://xiph.org/video/vid2.sh...

First off, that really only gives you 2 samples per WAVEFORM at 20 kHz.

Watch the video. It demonstrates how 2 samples is enough.

And the minimum phase difference in a CD corresponds to roughly a quarter inch difference in the free air path of the sound (napkin math) .

There's no loss of phase information in a quantized bandwidth-limited signal.

See this video https://xiph.org/video/vid2.sh... at 21:00

Disagree? Please watch this several times before hitting 'reply':
https://xiph.org/video/vid2.sh... [xiph.org]

Very interesting video. In particular the explanation of the stair-stepping stuff and how sampling actually works. Thanks!

+several million, informative.

a) If vinyl is "warmer" then that's just distortion
b) 44.1kHz, 16bits is absolutely enough for reproduction. There may be a case for using 48kHz to help with making real-world reconstruction filters but that's it. You absolutely do not need more than that for listening.

Disagree? Please watch this several times before hitting 'reply':
https://xiph.org/video/vid2.sh...

The revolt against GIF that culminated in PNG over 20 years ago showed that royalty free and patent unencumbered works. I recently learned of Free Lossless Image Format (FLIF), currently in development and intended to replace PNG, and it looks great. Handily beats WEBP, and also addresses animations, the reason GIF didn't completely die. FLIF alone shows that PNG was not a one time freak. But there's also Xiph, which has been working on audio and video codecs for decades, giving us Ogg Vorbis and fairly recently, Opus. And soon we will have AV1.

Opus supposedly bridges the gap in audio between music and voice, better at both than even the best codecs tuned specifically for one or the other. Opus is good at voice, but from my own experiences, no, it's not the very best. Shouldn't VoIP software use Opus, if it was the best? I'd love to abandon Skype. This is the first I've heard of Codec2, thanks for mentioning it.

The area most in need of an update is lossy stills, where JPEG is still supreme. (Why didn't JPEG 2000 catch on?) I saw a comparison on stills between AV1 and JPEG, and in my opinion, JPEG is clearly superior. I hope AV1 improves there, but I wonder if we could have a "double" JPEG, bump the 8x8 blocks up to 16x16, just as a stopgap?

Yet despite all this evidence, IP rights holders still refuse to concede that unencumbered works. What is accomplished by moaning the clearly wrong idea that codecs will not be developed? They really think they can persuade some deep pocketed organizations to swallow that propaganda and join with them? I don't follow that notion of "fractional options", and I don't see why anyone else should either, especially with AV1 in the wings. Meantime, for video I'm sticking with VP9 and Opus (and webVTT) inside WEBM.

AV1 outperforms JPEG. AV1 delivers a smaller file size at the same quality or better quality at the same file size. Try this comparison of JPEG and AV1 at the same file size.

Sell me pure 24 bit by at least 96 audio files..

I was 100% with you up until this point. 24-bit audio is great for mastering, but practically useless on the consumer end. The noise floor of 16-bit audio is already so low you can hear a mouse fart. As for the 96khz sampling rate, I'm sure you're one of those stupid ass motherfuckers that believe D/A converters produce "stair-step" waveforms. You're wrong and they don't. At 44.1, every frequency at and below 22,050hz is reproduced perfectly, smoothly, without any temporal distortion whatsoever. You don't understand how digital sampling works. Educate yourself.

The first link talks about 24/192 which is actually not positive - https://people.xiph.org/~xiphm...

Not OP but I can easily discern the harmonic beat between ultrasonic and sonic combinations.

In practice :
- actually not at all.

the "beat" frequency is calculated by the delta of the 2 frequencies.
to be able to hear the "beat" you'd need a frequency that is clearly in the hearing domain.
to make it beat with an inaudible ultra-sound you would need a frequency that is clearly above the hearing domain.
(i.e.: We're not speaking 10'000 Hz vs 10'001 Hz)

a delta between such two frequencies is sure to gives a beat frequencies that is beyond the response time of the ear.
i.e.: your inner ear labyrinth receptor won't be able to notice a sound going on and off that fast.

not to mention that the delta might end up being higher than the lower frequency at which point there's no really beat to be heard at all.
(a beat frequency is usuabl <<sound frequency)

in theory :
a beat is just a sound oscillating in volume over time (a type of tremolo if you wish).
in this case you're not actually hearing utlra sound (you don't have receptors for that), you're hearing distorsions in the audible domain, for which you DO have a receptor.
and similarly a sampling rate of 48kHz won't be able to code the ultra sound frequency (beyond nyquist frequency), but can clearly code an audible frequency whose amplitude varies over time.

but that's just about the theory of perceiving or recording a beat a.k.a. an osciliation of volume a.k.a. a tremolo.
for the practicality of a setup with a beat between an audible and inaudible frequency, see above.

for more informations :
- watch this video
- read this wiki

Not OP but I can easily discern the harmonic beat between ultrasonic and sonic combinations.

In practice :
- actually not at all.

the "beat" frequency is calculated by the delta of the 2 frequencies.
to be able to hear the "beat" you'd need a frequency that is clearly in the hearing domain.
to make it beat with an inaudible ultra-sound you would need a frequency that is clearly above the hearing domain.
(i.e.: We're not speaking 10'000 Hz vs 10'001 Hz)

a delta between such two frequencies is sure to gives a beat frequencies that is beyond the response time of the ear.
i.e.: your inner ear labyrinth receptor won't be able to notice a sound going on and off that fast.

not to mention that the delta might end up being higher than the lower frequency at which point there's no really beat to be heard at all.
(a beat frequency is usuabl <<sound frequency)

in theory :
a beat is just a sound oscillating in volume over time (a type of tremolo if you wish).
in this case you're not actually hearing utlra sound (you don't have receptors for that), you're hearing distorsions in the audible domain, for which you DO have a receptor.
and similarly a sampling rate of 48kHz won't be able to code the ultra sound frequency (beyond nyquist frequency), but can clearly code an audible frequency whose amplitude varies over time.

but that's just about the theory of perceiving or recording a beat a.k.a. an osciliation of volume a.k.a. a tremolo.
for the practicality of a setup with a beat between an audible and inaudible frequency, see above.

for more informations :
- watch this video
- read this wiki

once you lose the quality from the digitization process it is lost for good.

Lose what quality? 44.1KHz/16bit (CD quality) is way beyond what LP, reel-to-reel or even the much-vaunted master tapes can manage.

It sounds to me like you don't understand digital audio at all. You should watch this video: https://xiph.org/video/vid2.sh...

> Bullshit. H.265 is a superior encoding method,

[Citation needed]

I encourage you to head over to http://planet.xiph.org/ and read some of Monty's blog posts and links to double blind tests at various bit rates.

Even then the world is not much concerned with +/- 5% in subjective viewing tests. See Betamax vs. VHS or Vorbis vs. MP3. First to market, the licensing fun you pointed out, and marketing campaigns often matter more. If we were talking H.265 vs. VP8 you'd have a valid point, but VP9 and H.265 are close enough that a minor win or loss in a given encoding test is neither here nor there.

Unfortunately, even the best 16b-bit 44kHz reproduction chains introduce uncorrelated high-order harmonics that fall in the audible range

That old saw? Modern DACs reduce that distortion to inaudible levels at 16-bit, and with proper dithering it vanishes completely into the noise floor.

https://wiki.xiph.org/Videos/Digital_Show_and_Tell#Dither

The included license says "Attribution-NonCommercial-NoDerivatives 4.0 International"

https://media.xiph.org/video/d...

How can competitors use this if noncommercial clause attached?

For those interested, I recently setup a comparison between different image formats, including WebP, Daala, BPG and JPEG.

BPG is not viable due to the licensing situation around HEVC. Don't waste your time on formats which require patent royalty payments. AV1 is the future of web video, so a new still image format based on that (similar to WebP) is a better option.

These two videos are an excellent introduction to audio signal theory.

Highly recommended to all readers enjoying this thread.

No, Digital to Analog converts do not output square waves.
Nor does MP3 encoding or anything else.

Go learn a thing or two over here: https://www.xiph.org/video/

Except that anything at 20 kHz will be a triangular wave when sampled down to 44.1 kHz for CD-pressing.

No, it won't. 20kHz will be sampled and reproduced perfectly because the signal is over-sampled, a sharp low-pass filter applied, and the resulting signal will not contain any frequency content higher than 22.05 kHz - which includes any harmonics that might be part of your "triangular wave."

Don't believe me? You can see for yourself in this video at the 5:40 mark, where Monty shows how a 20kHz frequency is reproduced perfectly using a 44.1kHz sampling rate.

It's worth your while to watch the whole video. His signal generator and oscilloscope are both analog.

Do you even know how CD's work? They are digital. They provide a discrete time-based signal of audio amplitude.

If you were in my class, I would fail you. If you were my grad student, I would cut your funding.

Take a 44.1 kHz digital sample of a 20 kHz signal. Route that signal through a "true" D/A converter to speakers (or an oscilloscope), and you will get a triangular wave, as I said before, albeit with some 'walk' due to the introduction of false harmonics to the signal. That is, phase non-uniformity == false signals beyond the Nyquist limit of the sampling rate. A low-pass filter at 44.1 kHz might clean it up, but that would be cheating. A Fourier Transform (power spectrum of the signal) would show the false harmonics, and in far greater intensity than that shown in your video. The analog oscilloscope is set to 'lock-in' mode, to represent the resultant analog wave-form based on the signal from the D/A converter that the presenter used. (It might be performing a boxcar averaging of the signal, as the modulations of intensity are quicker than the human eye can detect. I didn't bother looking at the dial settings.) A strict D/A conversion would not yield a sinusoid.

Your vaunted "oversampling" D/A conversion is in reality a fancy marketing term for "signal interpolation, fitting the signal to a sine-type function", done continuously based on some number of samples at any given moment (up to 12,000 data points) –a moving fit to a sinusoid. This 'error correction' is called CIRC.

Go back to your own example video and look at the plot shown at 7:20. Do you see where the data-points are? Do you see the sine-function that is fitted to that data stream? That is your "oversampling" in action. You see, because a CD player 'knows' that it is reproducing an audio signal, it interpolates the data-stream to fit the signal to an intensity-modulating sinusoidal wave-form. More accurately, it was the engineers who designed the Red Book specs, and dictated the specs for the CODEC, meaning here the D/A conversion of the binary EFM-encoded data on a CD to an audio signal that you get when you play a 'music' CD.

Not all signals in the world are sinusoidal, you know. This is why good scientific papers plot only data-points, or sometimes draw in a spline, while making sure to label it as "a guide to the eye." Without such notation in the caption, a paper showing a Fig. like that at 7:20 would be rejected from publication in any respectable Journal. If a manuscript's context is audio signal-processing, then there are reasonable assumptions that can be made about a wave-form that is digitally under-sampled.

Rule of thumb: You need at minimum 10 data-points per period to be able to faithfully analyze a wave-form. This applies to audio, images with periodic features, and everything else.

Except that anything at 20 kHz will be a triangular wave when sampled down to 44.1 kHz for CD-pressing.

No, it won't. 20kHz will be sampled and reproduced perfectly because the signal is oversampled, a sharp low-pass filter applied, and the resulting signal will not contain any frequency content higher than 22.05 kHz - which includes any harmonics that might be part of your "triangular wave."

Don't believe me? You can see for yourself in this video at the 5:40 mark, where Monty shows how a 20kHz frequency is reproduced perfectly using a 44.1kHz sampling rate.

It's worth your while to watch the whole video. His signal generator and oscilloscope are both analog.

Devices are programmable. But devices (as opposed to computers) often have relatively wimpy CPUs and depend on specialized decoding hardware for acceptable performance. Switching from a codec that has hardware support to one that does not will not work well on such devices. That includes most set-top boxes and smart TVs.

Mobile has the same problem on older devices but a different one on new devices.The CPU might have enough computing power to handle the new codec but doing the decoding in the CPU rather than the video hardware is likely to shorten battery life considerably.

So I think my basic point stands. Switching to H.265 is not viable for the majority of the installed base right now, but if Netflix is willing to do dual encoding they could get some benefit by using it for the users that can handle it. That will change with time as more of the installed base gets H.265 hardware decoding. But perhaps by then the hot topic will be whether to switch to Daala: https://xiph.org/daala/

Meanwhile, if they have found a way to improve the efficiency of their H.264 encoding that's a win all around. Lower bandwidth bills for Netflix, less congestion on the internet pipes for everybody, and everybody's systems still work. And they plan to do the re-encoding by using their existing hardware during off-peak hours so the only expense is a bit of extra electricity, because modern computers consume more power when they're actually doing something than when they are idle.

If they actually want to support open codecs

Microsoft supports Opus because they have IP in it via their purchase of Skype. And Microsoft has joined the Alliance for Open Media to participate in the development of the video codec to follow VP9, which will be built from the best features of Thor, Daala, VP10 and whatever anyone else brings to the table.

Daala is dead. Long live NetVC. As Jean-Marc Valin said in the comments of the Cisco blog post: "The final NetVC codec will be neither Thor, nor Daala. It will be some kind of mix of the various contributions received. (disclosure: I'm in the Daala team at Mozilla)"

And as Timothy Terriberry said in an HN comment: "Hello, I'm the Daala tech lead. One of the things that made Opus a success was the contributions of others. We certainly don't have a monopoly on good ideas. We'll take pieces of Daala and stick them in Thor and pieces of Thor and stick them in Daala, and figure out what works best."

Theora, based on VP3, is roughly H.263-class technology comparable to Sorenson Spark (FLV) and MPEG-4 ASP (DivX and Xvid). H.264 and VP8 are a generation ahead of it in rate/distortion performance at Internet bitrates

That's mostly true, although Theora has been improved to the point where it is closer to H.264.

That said, Xiph has also been working on Daala, which is intended to compete with H.265.

The Daala team has also experimented with integrating some Thor's features into Daala. It's likely that the codec developed by the IETF Internet Video Codec working group will be built from the best features of Daala, Thor and any additional contributions.

create a competing standard that is designed specifically to avoid patents, and license it royalty-free

That's exactly what Xiph does with the Daala project. They're trying to implement lapped transforms for video (more or less the same principle as Opus does for audio) and since it's not based on traditional block encoding, Daala should avoid most patents. Their demos are already pretty impressive.

I do hope they have learned from Google's mistakes with VP9 and and come out early with a good specification and stable hardware reference implementation.

I work for a fairly large video oriented service, and we would love to start supporting alternative codecs (and eventually leave H264/5) but lack of full hardware support in mobile devices is an absolute blocker. Everything needs to be working well, with battery efficiency, on mobile devices at this point. I know a lot of other services currently on the fence about VP9 have similar views.

Google have been approaching video codecs as they approach online services - betas in production, frequent revisions, not a stable spec (or spec at all). Not a good approach to get the chip makers onboard.

I'll just leave this here: https://wiki.xiph.org/Daala

Better: Work together with like-minded companies to create a competing standard that is designed specifically to avoid patents, and license it royalty-free.

And better yet, do that work in the IETF's Internet Video Codec working group, which is what Xiph and Cisco are doing.

create a competing standard that is designed specifically to avoid patents, and license it royalty-free

That's exactly what Xiph does with the Daala project. They're trying to implement lapped transforms for video (more or less the same principle as Opus does for audio) and since it's not based on traditional block encoding, Daala should avoid most patents. Their demos are already pretty impressive.

Most studio microphone frequency charts typically drop off before or near 20kHz anyhow, so it's unlikely it would even be captured in the first place.

Pfft. Not that I care if people want to blow their money on formats or equipment that's over-engineered by several factors beyond what they could possibly hear. And if they feel a bit more special believing that, unlike most other humans, they alone have "golden ears" that can hear the difference... well, that's fine with me too. Just don't try to shovel that shit in my direction. Prove it to me with a blind A/B test, and then I'll take your claims seriously.

It's pretty telling when you actually hear what Neil Young thinks about compressed audio file formats:

“We’re in the 21st century and we have the worst sound that we’ve ever had. It’s worse than a 78 [rpm record]. What happened?

        “The MP3 only has 5 percent of the data present in the original recording The convenience of the digital age has forced people to choose between quality and convenience, but they shouldn’t have to make that choice.”

        “If you’re an artist and you created something and you knew the master was 100 percent great, but the consumer got 5 percent, would you be feeling good? “

It's clear he doesn't really understand the technology, and thinks that compressing a song to 1/20th of the original size means that it's only 5% of the value of the original. Yes, you can overcompress audio until it sounds like crap, and MP3 is getting a bit long in the tooth. That's why most people have switched to 256kbps AAC (Apple music streams at this quality, btw), and the overwhelming majority of people in A/B tests can't tell the difference between compressed and non-compressed audio, nor between 16-bit/44.1kHz vs high resolution audio.

There doesn't need to be a term for it, you can just be someone that appreciates good playback equipment and production in music. The word "audiophile" is now thoroughly linked with the kind of idiots you described. That kind of stench can never be thoroughly erased.

The incredulous claims about vinyl and cabling have been around for a long time, but more recently computers have opened up a whole new level of foolery. There are programs out there that supposedly "keep your audio data in the CPU cache" instead of RAM and have some BS rationale as to why that improves audio quality. Then there's the high-res music stuff that's been gaining particular traction in recent years (not to be confused with lossless compression which has gained traction around the same period) especially as "respected" musicians have come to launch placebo-based product lines for people who want their music oversampled to several times beyond the limit of human hearing. I keep this link handy as a particularly thorough rebuttal to the claims they make:
http://xiph.org/~xiphmont/demo...
The other nice thing about this link is it's from xiph.org - you know the guys who developed FLAC, Ogg Vorbis, Opus, etc.? They know more about audio reproduction, particularly digital audio reproduction, than Neil Young will ever know. He lacks either the will or the mental capacity to educate himself on the shit that's coming out of his own mouth when he gets up on his "digital music is terrible" soapbox. It's like, even if they accept that etched PVC disks aren't the pinnacle of audio reproduction (welcome to 1940), they have to invent some convoluted, cumbersome, expensive way to do digital audio so that their playback system is esoteric enough to please the Gods of Rock. I think Neil takes a particular shine to the "expensive" aspect, since he stands to profit from it all. But hey I'm sure he's a renown philanthropist or something, plus remember he wrote a couple decent songs 40 years ago? Fuck Neil Young.

Anything higher than 44.1/16 is scientifically proven to be wasteful.

http://xiph.org/~xiphmont/demo...

The BjÃntegaard metric is used to calculate the bitrate saving achieved by the test
encoders, based on the PSNR scores.

The problem with a lot of these studies is that the metrics don't always work that well. For example, look at the image comparison on pages 26, 27 and 28 in the NetVC presentation. The first codec on page 27 has a better PSNR score than Daala on page 28, yet to me the image compressed by Daala looks better and has more detail.

Daala's not ready yet but it's been proposed as the basis for the NetVC implementation. NetVC will probably end up being Daala merged with other contributions.

My guess is that VP9 probably isn't quite as good as H.265, but it is definitely in the ballpark.

You'd be wrong about that actually. Monty's given it his usual expert and honest analysis, see one of his blog posts from late last year. Caveat: If you compare VP9 today vs. some tuned H.265 of the future the roles may reverse. Or not. Who knows that's just pure speculation and it's not like VP9 won't tune up either.

But VP9 is not too late for the war with H.265

In fact VP9 spec was finalized quarters before H.265, and Google has the ear and other anatomical bits of all the hardware manufactures in the Android world, so VP9 hardware support from the start is in very good shape.

And what is never mentioned in the press releases is that VP9 and H.265 make their impressive bandwidth (or filesize) improvements at the cost of double the CPU needs. You do not want to be running these codecs without hardware support.

The exciting stuff is Daala.

What comes next? H.266?

Hopefully it will be Daala.

May Daala save us all.

> So not able to get into a discussion of recording of the music as time has progressed, just the fact it's sampled and not a true wave form.

YES, it IS a "true wave form." The standard link people post now on this topic is Monty Montgomery's (xiph) video "Digital Show and Tell." Watch and learn...digital audio fully captures the original signal with better fidelity than any vinyl: http://xiph.org/video/vid2.shtml

Two points is a "saw-tooth" wave. You cannot get a true sine wave out of a DAC, no matter how many trillions of samples you take as there are still a fixed number of values (12bit, 16bit, etc.) to measure the signal.

Jesus, you literally do not know what you are talking about. Watch Monty's video and you will see that you are dead wrong. Digital can faithfully reproduce any sine wave you throw at it as long as it's lower than 1/2 the sampling frequency.