Hi resolution audio and extended treble response!

[voodooless]

Like I have said later sampling rate you are taking 192000 point per second to capture at 20Hz cycle waveform.

Yeah, but so what? Why is this important?

 

[kongwee]

Yeah, but so what? Why is this important?

I dunno what to say. When 24/192k or DSD were presented, we were assume we no longer heard different in details in 20-20kHz if we push for more sampling rate. In analogue, 16/44.1 is Retina screen in ASR while other say it is 24/192 or DSD is Retina. Retina is how many pixel points in an square inch before we can't differential the image quality, just like how many points can we push in 1 sec of sample. If you can't understand, you have to read how PCM coding works. You will see example they are using points as example. You can have many version of the same thing in google.

 

[Killingbeans]

Like I have said later sampling rate you are taking 192000 point per second to capture at 20Hz cycle waveform.

Doesn't matter. You get exactly the same waveform. There's no "distortion". Watch the Monty video again, and really try to understand it this time.

Don't be this guy:

 

[antcollinet]

You can have 192,000 point to draw 20Hz sinewave instead of 44100 point. That what the Sony is talking about on the purple graph.

It doesn't matter. You can reproduce 20 to 20KHz perfectly with 44100 samples per second. You can't make it "more perfect" by adding more samples - all you can do is extend the frequency range byond 20Khz which is a waste of effort since we cannot hear it.

 

[voodooless]

If you can't understand, you have to read how PCM coding works

It's clear that you don't understand the sampling theorem

 

[kongwee]

Doesn't matter. You get exactly the same waveform. There's no "distortion". Watch the Monty video again, and really try to understand it this time.

If it just a simple sinewave in 20 cycle per second, no difference. If each cycle you have a sort have wave, like 20000 of square, saw waves in positive and negative per cycle. You have it in 20 different cycle per second. That is a complex harmonics in that 20 cycle per second. You need have more points to capture all the shape in those 20 cycle per second. Of course, if you can't heard the harmonics, it is just a simple sinewave 20Hz hum.

It's clear that you don't understand the sampling theorem

I know, it exist. It state you can accurately recover a signal to sample the frequency at 2 times. That is theorem. 60hz screen refreshing rate sure be the max we can differential. However, 120hz refresh screen rate exist in gaming monitor and phone, we still can differential.

 

[Killingbeans]

You don't get it.

If the sample rate is 2x the frequency or more, you litterally get a perfect waveform. Doesn't matter how many more samples you add.

There are no stairsteps when you use a correct reconstuction filter. It seems like magic, but it's not.

 

[antcollinet]

If it just a simple sinewave in 20 cycle per second, no difference. If each cycle you have a sort have wave, like 20000 of square, saw waves in positive and negative per cycle. You have it in 20 different cycle per second. That is a complex harmonics in that 20 cycle per second. You need have more points to capture all the shape in those 20 cycle per second. Of course, if you can't heard the harmonics, it is just a simple sinewave 20Hz hum.

but the harmonics to create that wave shape can only be heard if they are lower than 20KHz.

So any that are above 20Khz can be filtered out without affecting what we hear. The ones below 20KHz can be reproduced perfectly with 44.1KHz sampling. So again - more is not needed.

 

[kongwee]

but the harmonics to create that wave shape can only be heard if they are lower than 20KHz.

So any that are above 20Khz can be filtered out without affecting what we hear. The ones below 20KHz can be reproduced perfectly with 44.1KHz sampling. So again - more is not needed.

If we filter out above 20kHz harmonic in 20Hz cycle, we are altering the signal. That not what sampling theorem taught us. lol

 

[Killingbeans]

I'm sorry, but you're wrong. Nyquist–Shannon requires a finite bandwidth.

Think about what a "stairstep" actually is. It can only exist as frequency content above the intended signal. When you filter it out, you don't lose information. You just remove garbage that doesn't belong to the original signal.

 

[voodooless]

lol

... indeed

 

[kongwee]

I'm sorry, but you're wrong. Nyquist–Shannon requires a finite bandwidth.

Think about what a "stairstep" actually is. It can only exist as frequency content above the intended signal. When you filter it out, you don't lose information. You just remove garbage that doesn't belong to the original signal.

Those garbages are the timbre of the 20Hz cycle. it will never the same signal. It is not a simple add and subtract stuff. lol

 

[voodooless]

Those garbages are the timbre of the 20Hz cycle. it will never the same signal. It is not a simple add and subtract stuff. lol

Does 440 nm blue get less blue to you if you add ultraviolet?

 

[antcollinet]

If we filter out above 20kHz harmonic in 20Hz cycle, we are altering the signal. That not what sampling theorem taught us. lol

You filter it out before sampling.

And it is EXACTLY what sampling theorem taught us. The sampling theorem is always about a band limited signal. That is why 44.1kHz is chosen. You band limit to 20KHz (limit of human hearing), then you can capture everything with a 40KHz sample rate. That is increased slightly to 44.1kHz to allow space for an imperfect filter.

If you made any attempt to understand the Montey video, you'd get all this.

 

[antcollinet]

Those garbages are the timbre of the 20Hz cycle. it will never the same signal. It is not a simple add and subtract stuff. lol

It is the parts of the timbre that human hearing cannot detect. You ears are also band limited: they will also filter out anything above 20kHz (probably much lower if you are not a teenager who's never attended a rock concert).

 

[BDWoody]

If we filter out above 20kHz harmonic in 20Hz cycle, we are altering the signal. That not what sampling theorem taught us. lol

Maybe you should start with understanding sampling theory a bit better before you keep laughing.

That's enough for this thread.

 

[maxaudio]

I am currently writing a paper about music streaming services and comparing them to Hi-Res audio files. I found an article about Hi-Res very informative (https://www.mobilefidelity-magazin.de/macht-den-hi-res-schwindel-sichtbar-xivero-musicscope) in German. In the meantime, I have also analyzed my files with various tools such as Musicscope. However, I have a question that is not entirely clear to me. As shown in my images and in the informative YouTube video by Audio Science Review, described at minute 9 (

), I would be interested to know if this high shelf boost comes from noise shaping or from a limiter artificially boosting the upper frequency spectrum, similar to a high shelf EQ. In an article on Audio Science Review, this "high shelf" or "boost" is referred to as "noise shaping".

I have attached an image that I used from HD downloads in PCM format for my paper. I would appreciate any help so that I can describe this correctly. Best Regards Max

 

[RandomEar]

I am currently writing a paper about music streaming services and comparing them to Hi-Res audio files. I found an article about Hi-Res very informative (https://www.mobilefidelity-magazin.de/macht-den-hi-res-schwindel-sichtbar-xivero-musicscope) in German. In the meantime, I have also analyzed my files with various tools such as Musicscope. However, I have a question that is not entirely clear to me. As shown in my images and in the informative YouTube video by Audio Science Review, described at minute 9 (

), I would be interested to know if this high shelf boost comes from noise shaping or from a limiter artificially boosting the upper frequency spectrum, similar to a high shelf EQ. In an article on Audio Science Review, this "high shelf" or "boost" is referred to as "noise shaping".

I have attached an image that I used from HD downloads in PCM format for my paper. I would appreciate any help so that I can describe this correctly. Best Regards Max

You mean the hump above 50 kHz in the screencap and the video? Yeah, that 100% looks like noise shaping. Amir also quickly explains it in the video you linked after minute 9. I don't quite understand what you refer to when talking about a limiter or high shelf - the signal in that region is definitely not intentionally added into the files.

Also, quick tip: You can link to timestamps in youtube video by pressing "Share" and then clicking the checkbox in front of "Start at [X:YY]"

 

[maxaudio]

You mean the hump above 50 kHz in the screencap and the video? Yeah, that 100% looks like noise shaping. Amir also quickly explains it in the video you linked after minute 9. I don't quite understand what you refer to when talking about a limiter or high shelf - the signal in that region is definitely not intentionally added into the files.

Also, quick tip: You can link to timestamps in youtube video by pressing "Share" and then clicking the checkbox in front of "Start at [X:YY]"

Hi randomear, thank you very much for the tip. I still copy and paste the address bar. That's very old school, I need to change that. I was a bit unsure because of the German article https://www.mobilefidelity-magazin.de/macht-den-hi-res-schwindel-sichtbar-xivero-musicscope. Let me translate this part:


"Unfortunately, not all files that we put under MusicScope were flawless. For example, a 176.4 kHz / 24-bit AIFF file of Michael Jackson's "Thriller" had always seemed suspicious to us because the sound was really lacking for a Hi-Res recording. Reading it into MusicScope supported this suspicion:
XiVero MusicScope Bitmonitor The Bit Monitor shows that only 16 bits are used out of the specified 24 bits in "Thriller". It appears that the upper harmonics spectrum has been slightly extended with the use of a maximizer, which MusicScope's automatic detection does not understand (hence the unusual threshold of 35.1 kHz, which would mean a sampling rate of 70.2 kHz), but which suggests the very clear step at about 22 kHz, i.e., the cutoff frequency of a CD. To fill the rest of the frequency spectrum up to 176.4 kHz, the file seems to have been initially converted to DSD, which shows the typical increase outside the audible range, and then converted back to PCM. Looking at the Bit Monitor shows that only 16 of the 24 bits are effectively used. It appears that clever forgers were indeed at work here."

This is why I'm asking for advice here in the forum. Amir's video was very plausible to me. I will describe my Screenshots as noise shaping.
Thanks

 

[voodooless]

Several things:
- If you sample at 176.4 kHz, then the spectrum only goes to half that. Error in the article…
- Why would a PCM -> DSD -> PCM conversion only yield 16 bits?