[roach7]

Can someone explain how upscaling a lossy digital signal makes the sound better? Wouldn't it just magnify the lossy signal? I don't understand how this works...

 

[danadam]

Before starting speculating on "how", it would prudent to first verify if it does make the sound better (whatever "better" means). And even before that, if it makes any audible difference at all.

 

[roach7]

Before starting speculating on "how", it would prudent to first verify if it does make the sound better (whatever "better" means). And even before that, if it makes any audible difference at all.

that's exactly what i'm trying to figure out but from reading on the m-scaler thread many people claim they hear significant difference with and without the m-scaler

 

[bigshot]

You can't go by anecdotal comments in internet forums. People claim to hear differences in all kinds of things that have no audible differences. The way to tell is with a blind A/B test.

If the difference is so small that it takes controlled testing to know if it even exists, odds are it really doesn't matter. Especially if the way it works involves extra processes that shouldn't make a lick of difference. This isn't the sort of thing I would normally think twice about.

 

[Audi5000]

if its a digital file, you cant make it better by changing the sample rate.. what you CAN do is take the relevant data excite it a bit. and tone down the noise/useless data..

its like taking a crappy jpeg image and adding slight blur and then changing tones and contrast.. and somehow its more pleasing to the eyes. but in bits and bytes, totally different.

or in short, sometimes blurring the background gives a false sense of the foreground being more clear.

 

[roach7]

so i'm guessing the extra bits added somehow "enhances" or "colors" the audio signal to a point people can hear a difference... i guess i'll have to a/b myself one day to really see. but not sure i'm going to plunk down 5k to see...

 

[bigshot]

Is this a DSP or is it simply upping the sample rate? "Exciting the data" sounds like complete hogwash to me. If there is no actual signal processing going on- just upscaling, my bet is that it makes absolutely no audible difference. They just include the feature for marketing and advertising purposes to fool people who don't know any better.

Upscaling doesn't change sound quality. Signal processing does.

 

[VNandor]

Can someone explain how upscaling a lossy digital signal makes the sound better? Wouldn't it just magnify the lossy signal? I don't understand how this works...

What makes you think the digital signal is "lossy" to begin with? Do you have any understanding on how digital signals work in general? If not, you won't understand what the chord m-scaler is doing. Here's a really good video that should bring you up to speed with the fundamentals of digital audio.

Is this a DSP or is it simply upping the sample rate?

It only does upsampling.

 

[Lucky87]

Here is a very good explanation of it here

 

[roach7]

thanks all for the links... will be checking them out to learn more!

 

[Audi5000]

Is this a DSP or is it simply upping the sample rate? "Exciting the data" sounds like complete hogwash to me. If there is no actual signal processing going on- just upscaling, my bet is that it makes absolutely no audible difference. They just include the feature for marketing and advertising purposes to fool people who don't know any better.

Upscaling doesn't change sound quality. Signal processing does.

i didnt know its not processing anything or adding anything, but in that case not worth the price at all imho.

i dont know what the native "upscaler" does on my samsung music app.. but it appears to do a little bit of dsp'ing.. seems to do a little bit of stereo expansion and treble boosting.. in digital audio workstations the stereo expansion is one of my favorite vst plug ins, but i do everything in 16bit 44.1 wav's. i have some 24 bit wav's and they dont seem to be affected by my phone's upscaler.

 

[bigshot]

It only does upsampling.

Then it's as useless as teats on a bull hog.

 

[mammal]

Can someone explain how upscaling a lossy digital signal makes the sound better? Wouldn't it just magnify the lossy signal? I don't understand how this works...

I wondered about this myself in the past, so much so that I tried HQPlayer as well as M Scaler connected to my Hugo TT 2. In both cases, I was trying very hard to hear a difference, did a blind ABX test with my wife and I managed to accurately tell which one is which. Did I actually prefer the "upsampled"? Funnily enough, I did not. I am in this hobby for enjoyment, trying gear left and right, in order to find something that sounds good to my ears. NOT necessarily accurate, as I listen mainly to EDM, so who the heck knows how that is "really" supposed to sound.

They way PCM and value-hold DACs were explained to me in the past was that PCM at 44.1 is a series of discrete (digital) samples of a recorded continuous (analogue) sound. Based on the math of the famous Nyquist–Shannon sampling theorem, as long as you sample 2x of the highest band limited frequency, you will be able to perfectly reconstruct the original analogue waveform, from a series of discrete values. Now as far as max frequency of your DAC goes, let it be 44.1 * 16 = 705.6 or 48 * 16 = 768, that would be the max resolution of the DAC, so that the input PCM file of said frequency on a value-hold DAC has the most detail possible. Why would you want to upsample from original input of 44.1? So that the jumps between samples (if you use value-hold aka zero order-hold) method (no interpolation/approximation between samples) so that you do not have jumps over less resolution (44.1) but more resolution (705). In essence, it is putting in new discrete values, so that jumps between samples are not as high.

Nyquist–Shannon sampling theorem has one requirement, that is quite difficult to achieve in practice - it requires "a sinc function, where these sinc functions are summed into a continuous function. A mathematically equivalent method is to convolve one sinc function with a series of Dirac delta pulses, weighted by the sample values. Neither method is numerically practical. Instead, some type of approximation of the sinc functions, finite in length, is used. The imperfections attributable to the approximation are known as interpolation error." I believe this is what M Scaler tries to do, it uses better interpolation, they call it number of taps, as infinite is not possible, more taps you have, better the result in practice.

You asked for a laymen's explanation, I did my best to provide it, I hope I did not butcher it completely. Not sure how much of this is audible, but my understanding is that if you had infinite sinc (or diract pulses) you would be able to reconstruct perfectly. Again, worth repeating is that I tried both HQPlayer and MScaler with a blind ABX testing, and I did not like what it did to my music.

 

[71 dB]

They way PCM and value-hold DACs were explained to me in the past was that PCM at 44.1 is a series of discrete (digital) samples of a recorded continuous (analogue) sound. Based on the math of the famous Nyquist–Shannon sampling theorem, as long as you sample 2x of the highest band limited frequency, you will be able to perfectly reconstruct the original analogue waveform, from a series of discrete values. Now as far as max frequency of your DAC goes, let it be 44.1 * 16 = 705.6 or 48 * 16 = 768, that would be the max resolution of the DAC, so that the input PCM file of said frequency on a value-hold DAC has the most detail possible. Why would you want to upsample from original input of 44.1? So that the jumps between samples (if you use value-hold aka zero order-hold) method (no interpolation/approximation between samples) so that you do not have jumps over less resolution (44.1) but more resolution (705). In essence, it is putting in new discrete values, so that jumps between samples are not as high.

Nyquist–Shannon sampling theorem has one requirement, that is quite difficult to achieve in practice - it requires "a sinc function, where these sinc functions are summed into a continuous function. A mathematically equivalent method is to convolve one sinc function with a series of Dirac delta pulses, weighted by the sample values. Neither method is numerically practical. Instead, some type of approximation of the sinc functions, finite in length, is used. The imperfections attributable to the approximation are known as interpolation error." I believe this is what M Scaler tries to do, it uses better interpolation, they call it number of taps, as infinite is not possible, more taps you have, better the result in practice.

You asked for a laymen's explanation, I did my best to provide it, I hope I did not butcher it completely. Not sure how much of this is audible, but my understanding is that if you had infinite sinc (or diract pulses) you would be able to reconstruct perfectly. Again, worth repeating is that I tried both HQPlayer and MScaler with a blind ABX testing, and I did not like what it did to my music.

There are no "jumps" between sample points in analog world. Jumps would require infinite bandwidth. DACs don't have infinite bandwidth. Amps don't have either. Speakers and headphones don't have either. Our ears filter effectively frequencies above 20 kHz. So, "jumps" are no more when the sound enters our ear. It is continuous analog signal. Adding smaller "jumps" with upsampling changes nothing.

I mentioned DACs don't have infinite bandwidth. That's why they only "try" to do that value-holding business. The voltage changes in finite time (determined by the time constant of the circuitry) from the previous sample value to the new one. This is where lower sample rates have an advantage. Smaller percentage of the "hold" sample values is wrong. The faster sample rate the more the circuitry struggles to "keep up" with the fast changing sample values. Manufacturers don't tell you this stuff, because they just want to sell you fancy stuff. You do not need $5k upsamplers to enjoy music. Sure, 40 years ago DACs had a lot room for improvement, but those days are over. Digital audio technology has matured since. Stunning performance is cheap, hundreds of dollars if even that rather than thousands of dollars. Save your money and invest it elsewhere were it has real impact such as improving the acoustics of your listening room or binaural processing for headphones.

The impulse response of a linear phase brick-wall filter is sinc function, which is infinitely long in time, but it is windowed into a finite length. This windowing causes the filter have its frequency response at the cut of frequency, which is nothing to write home about. Also, there is not need to use the default filter. There are other options for filtering and in my experience this has a very small impact on the spatial width of the sound: So it is possible to tinker with these things without upsampling to obscene sample rates.

Anyway, if someone becomes happy spending $5k on an upsampler that is not my money or my business. I would never ever do that myself, but then again it is difficult to have millionaire logic when you are not a millionaire, but a regular dude trying to pay the rent and bills.

 

[mammal]

There are no "jumps" between sample points in analog world. Jumps would require infinite bandwidth. DACs don't have infinite bandwidth. Amps don't have either. Speakers and headphones don't have either. Our ears filter effectively frequencies above 20 kHz. So, "jumps" are no more when the sound enters our ear. It is continuous analog signal. Adding smaller "jumps" with upsampling changes nothing.

I mentioned DACs don't have infinite bandwidth. That's why they only "try" to do that value-holding business. The voltage changes in finite time (determined by the time constant of the circuitry) from the previous sample value to the new one. This is where lower sample rates have an advantage. Smaller percentage of the "hold" sample values is wrong. The faster sample rate the more the circuitry struggles to "keep up" with the fast changing sample values. Manufacturers don't tell you this stuff, because they just want to sell you fancy stuff. You do not need $5k upsamplers to enjoy music. Sure, 40 years ago DACs had a lot room for improvement, but those days are over. Digital audio technology has matured since. Stunning performance is cheap, hundreds of dollars if even that rather than thousands of dollars. Save your money and invest it elsewhere were it has real impact such as improving the acoustics of your listening room or binaural processing for headphones.

The impulse response of a linear phase brick-wall filter is sinc function, which is infinitely long in time, but it is windowed into a finite length. This windowing causes the filter have its frequency response at the cut of frequency, which is nothing to write home about. Also, there is not need to use the default filter. There are other options for filtering and in my experience this has a very small impact on the spatial width of the sound: So it is possible to tinker with these things without upsampling to obscene sample rates.

Anyway, if someone becomes happy spending $5k on an upsampler that is not my money or my business. I would never ever do that myself, but then again it is difficult to have millionaire logic when you are not a millionaire, but a regular dude trying to pay the rent and bills.

Thank you for your message, I am here to learn. I did try Chord's upsampler for a week, did not like what it did to the music, so returned it. Since I do not have your level of understanding of these technologies, I am easy to sway away by a talk of a salesman / marketing department on what will improve my music enjoyment, especially if they quote math to me, which to my low level of understanding sounds reasonable. But in the end, I vote with my wallet.