Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: From 44.1 to 96, which is the best resampler? (Read 67810 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

From 44.1 to 96, which is the best resampler?

X-FI Titanium harware srs ?
srs ?
ssrs ?
pphs ?

From 44.1 to 96, which is the best resampler?

Reply #3 –
What are you trying to achieve by resampling?

From 44.1 to 96, which is the best resampler?

Reply #5 –
Quote
  • Upsampling to progressively higher sample rates makes progressive improvements to fine detail, sound stage
    depth and image separation. So, the sound quality increases as you upsample CD data first to 24/88.2, then 24/176.4, then 24/352.8 kS/s.
  • Converting 16/44.1 to 24/44.1 kS/s makes a worthwhile improvement to the fine detail, so the resolution is important also.


These claims lie purely in the realm of fantasy, magic and fairy tales. You can't make something from nothing. It's impossible to "restore" or otherwise add detail in a digital stream which isn't there.

From 44.1 to 96, which is the best resampler?

Reply #6 –
If your sound card upsamples 44.1 to 48 kHz, but does a poor job of it (as some do) then you can retain more quality by applying a good upsampler before playing your files. Either way there is some loss of quality, but some resamplers lose less quality than others.

From 44.1 to 96, which is the best resampler?

Reply #7 –
Quote from: JensRex on
Quote
  • Upsampling to progressively higher sample rates makes progressive improvements to fine detail, sound stage
    depth and image separation. So, the sound quality increases as you upsample CD data first to 24/88.2, then 24/176.4, then 24/352.8 kS/s.
  • Converting 16/44.1 to 24/44.1 kS/s makes a worthwhile improvement to the fine detail, so the resolution is important also.


These claims lie purely in the realm of fantasy, magic and fairy tales. You can't make something from nothing. It's impossible to "restore" or otherwise add detail in a digital stream which isn't there.

i dun think so
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic
But how "smooth" it is, I think it depends on the implemention of the resampling algor

Quote from: pdq on
If your sound card upsamples 44.1 to 48 kHz, but does a poor job of it (as some do) then you can retain more quality by applying a good upsampler before playing your files. Either way there is some loss of quality, but some resamplers lose less quality than others.

ya
I agree with you
so what i need is choosing a good real-time resampler from 44.1 to 96~ 
any evidence to prove that all of the above is not good resamplers to achieve this purpose ?

From 44.1 to 96, which is the best resampler?

Reply #8 –
Quote from: JoE YuNG * Metalphile on
i dun think so
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic


You are right about this, but smoothing certainly doesn't add any detail. Maybe you've seen some pseudo-scientific shows on TV like CSI where they take a 150*150 pixel picture of a license plate and blow it up to size "barn door", and then make out a child picking his nose in the background. This is analogous to what you're trying to do. You can't add detail to something which isn't there to begin with. Be that sound or images.

You can maybe achieve some sort of "effect" or otherwise, but it's not restoring anything.

I also have no idea what you mean by intersample zero data. This phenomenon only exists in your head.

From 44.1 to 96, which is the best resampler?

Reply #9 –
Quote from: JoE YuNG * Metalphile on
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic
But how "smooth" it is, I think it depends on the implemention of the resampling algor

“Smoothing” is what the low-pass filter (part of the DAC) does. Even without a lowpass, the artifacts would be in a frequency range outside the limits of most speakers and human ears. Additionally, they would have very low energy in almost every case (of course you could generate a 22050 Hz signal at full amplitude which would lead to strong harmonics, but natural sounds/music/speech don’t contain that).
FLAC.

From 44.1 to 96, which is the best resampler?

Reply #10 –
Quote from: JensRex on
Quote from: JoE YuNG * Metalphile on
i dun think so
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic


You are right about this, but smoothing certainly doesn't add any detail. Maybe you've seen some pseudo-scientific shows on TV like CSI where they take a 150*150 pixel picture of a license plate and blow it up to size "barn door", and then make out a child picking his nose in the background. This is analogous to what you're trying to do. You can't add detail to something which isn't there to begin with. Be that sound or images.

You can maybe achieve some sort of "effect" or otherwise, but it's not restoring anything.

I also have no idea what you mean by intersample zero data. This phenomenon only exists in your head.

what i mean is linear interpolation - adding data between 2 data points
Can't interpolation smooth the curve ?
http://en.wikipedia.org/wiki/Interpolation

doesn't upsampling use similar function ? 

From 44.1 to 96, which is the best resampler?

Reply #11 –
Quote from: JoE YuNG * Metalphile on
what i mean is linear interpolation - adding data between 2 data points
Can't interpolation smooth the curve ?
http://en.wikipedia.org/wiki/Interpolation

doesn't upsampling use similar function ? 
But the smoothing algorithm will almost certainly select the wrong* value for the new sample between the two data points.

*wrong in the sense that it is very unlikely that the smoothing algorithm will pick the value of the sample that would have been sampled had the audio been sampled at the new, higher sampling rate.

From 44.1 to 96, which is the best resampler?

Reply #12 –
i agree that upsampling = more faked details
However, more faked deta  <> worse sound 

From 44.1 to 96, which is the best resampler?

Reply #13 –
Quote from: JoE YuNG * Metalphile on
i dun think so
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic
But how "smooth" it is, I think it depends on the implemention of the resampling algor


As a graphics developer, I must say, you are utterly confused as to how antialiasing in graphics works.
With the current techniques, you either draw the scene in a higher resolution and scale it down to the display resolution (supersampling) or you take additional samples in different patterns (multisampling) in a neighborhood around each "pixel center", ending up with more coverage information to produce the final image with.

Common to both ways is that you never conjure information out of thin air, you just sample the source geometry more.

What you describe is closest to a simple blur, which by averaging together nearby fragments obliterates higher frequencies in the image to achieve a somewhat smoother result.

In order to do anything remotely resembling graphics antialiasing in audio, you need a signal with higher sampling rate and bit depth in order to sample and process it.
Stay sane, exile.

From 44.1 to 96, which is the best resampler?

Reply #14 –
Quote from: Zao on
Quote from: JoE YuNG * Metalphile on

i dun think so
I think resampling in audio is similar to anti-aliasing in graphic
It's replace the zero data between the samples in the signal with some calculated value.
The output signal therefore becomes "smoother"
like you can see the result when using anti-aliasing in graphic
But how "smooth" it is, I think it depends on the implemention of the resampling algor


As a graphics developer, I must say, you are utterly confused as to how antialiasing in graphics works.
With the current techniques, you either draw the scene in a higher resolution and scale it down to the display resolution (supersampling) or you take additional samples in different patterns (multisampling) in a neighborhood around each "pixel center", ending up with more coverage information to produce the final image with.

Common to both ways is that you never conjure information out of thin air, you just sample the source geometry more.
What you describe is closest to a simple blur, which by averaging together nearby fragments obliterates higher frequencies in the image to achieve a somewhat smoother result.

In order to do anything remotely resembling graphics antialiasing in audio, you need a signal with higher sampling rate and bit depth in order to sample and process it.


In fact, I don't know wt algorithm is used in resampling in audio 

and that's why i want to know if good resampler can really improve the sound.

if yes, can real-time resampler achieve this ?

if yes, which resampler is the best ?

From 44.1 to 96, which is the best resampler?

Reply #15 –
Many people who have soundcards that resample to 48kHz use the resampler in foobar2000 to software resample their 44100 material to 48000 simply because the software resampler does a better job than does their soundcard.

I've done loopback tests on the standard PPHS resampler in foorbar and even in standard mode (ultra mode not selected) the artifacts I measured were definitely way below audible (probably at least 30dB below audible). I have no hesitation in saying it would be transparent to anyone.

Resampling cant interpolate information that wasn't there to begin with, but neither does it interpolate any "wrong" details either. It just reconstructs the the signal in presicely the same way as your DAC should reconstruct it anyway (even if kept at 44100). The only way that software upsampling can improve sound is if your DAC is not really doing it's job properly. This senario is not entirely impossible so you might well wish to try upsampling and see if you can notice a difference. Just be aware that you (like everyone else) will be highly prone to imagining differences that aren't there unless you use "blind" tests. At a minimum I'd say you should get someone to change the resampling settings in foobar (selecting either "no resampling" or  96000) without you knowing which is selected (get them to toss a coin or something to select) and test if you can pick which is which. I'll bet you cant pick it accurately.

From 44.1 to 96, which is the best resampler?

Reply #16 –
@JoE YuNG * Metalphile:
Did you notice that in the URL you posted there is the word "myths" which exactly describes the scientific background of the advice .

From 44.1 to 96, which is the best resampler?

Reply #17 –
Quote from: JoE YuNG * Metalphile on
and that's why i want to know if good resampler can really improve the sound.

Resampling is processing. Processing is manipulation. With digital audio, manipulation is degradation 

If you were to upsample, upsample to 88.2 kHz or 176.4 kHz unless your chain doesn't support either sample rate. But don't upsample!

From 44.1 to 96, which is the best resampler?

Reply #18 –
Quote from: uart on
The only way that software upsampling can improve sound is if your DAC is not really doing its job properly.
Since most modern DAC's are using half-band filters it is very well possible that upsampling gives some (theoretical) benefits. Half-band filters have a -3dB response at the Nyquist frequency and therefore can not avoid aliasing. Upsampling to a double rate (88.2 or 96 kHz) will move the aliasing artifacts one octave higher, out of the audible range.
If aliasing or other low-pass filter artifacts are a problem (read: are audible) at the standard rate, upsampling might give some benefits.
Increased resolution or "smoothing" by upsampling sounds to me like marketing hogwash from an otherwise highly respected company.

Kees de Visser

From 44.1 to 96, which is the best resampler?

Reply #19 –
Quote from: Kees de Visser on
Since most modern DAC's are using half-band filters it is very well possible that upsampling gives some (theoretical) benefits.

Let me emphasis your "theoretical": Suppose your lowpass filter's transition band is from 20-24 kHz, with -6 dB at 22 kHz and -110 dB above 24 kHz. These image frequencies above 22 kHz don't hurt at all.

Quote from: Kees de Visser on
Half-band filters have a -3dB response at the Nyquist frequency and therefore can not avoid aliasing.

It's -6dB for half-band filters. And it's "imaging" what they are supposed to suppress -- not aliasing.

Cheers,
SG

From 44.1 to 96, which is the best resampler?

Reply #20 –
Quote from: Zao on
In order to do anything remotely resembling graphics antialiasing in audio, you need a signal with higher sampling rate and bit depth in order to sample and process it.


If compared with pictures, I would say that audio signal is plain bitmap picture of fixed resolution - you can't get bigger picture of better quality from a small picture. however, IF audio would be somehow "vectorised", you could blow it up to whatever resolution you want. But I don't know of any way to do that, and I doubt that it will ever appear.

  it would be nice, though... you have a stream of data, and reproduce it in whatever resolution you want.
TAPE LOADING ERROR

From 44.1 to 96, which is the best resampler?

Reply #21 –
Wikipedia says the process of decoding a Pulse-Density Modulation signal into an analog one is amazingly simple; does that mean it's easier to make a high-quality DSD DAC than to make a PCM DAC? Couldn't DSD put an end to hunting for the best DAC, and resampling?

From 44.1 to 96, which is the best resampler?

Reply #22 –
I don't see how as the final resolution is still determined by the sample-rate and the timing of the pulses is just as critical as the required accuracy in bit-significance in current DACs.

Any clock jitter would cause errors in amplitude accuracy at the output akin to frequency modulation techniques used for FM radio transmission, and any amplitude error in the pulses themselves just adds to this problem.

It seems, at first sight, to move a few problems into the digital domain where they're much harder to correct for than in the digital domain with the current DAC approach, IMO.

Cheers, Slipstreem. 

From 44.1 to 96, which is the best resampler?

Reply #23 –
As far as I know it is hard to build resistors as precise as needed in some types of DACs, especially for higher bit-depths, while it is relatively easy to generate a stable clock signal (which is needed anyway). Therefore many DACs are already "1-bit-DACs". According to the German website Tweakpc.de, the Realteck chip ALC889A (used on my mainboard ) uses a 1-Bit DAC at 64x44.1 kHz and produces very good to excellent results in RMAA (I don’t have an English source at hand).

Edit: typo
FLAC.

From 44.1 to 96, which is the best resampler?

Reply #24 –
Quote from: skamp on
Wikipedia says the process of decoding a Pulse-Density Modulation signal into an analog one is amazingly simple; does that mean it's easier to make a high-quality DSD DAC than to make a PCM DAC?

What's a high quality "DSD DAC"? What does it do better than a low quality "DSD DAC"? What's a "DSD DAC" supposed to do anyway? -- reconstruct faithfully all the high frequency noise that's present in the DSD signal? Obviously not.

The problem with DSD is actually the recording/production step. That, and that the format itself is highly inefficient. => 1 bit signals for audio is a very bad idea.

Quote from: skamp on
Couldn't DSD put an end to hunting for the best DAC, and resampling?

You make it sound like it's hard to get hold of a decent "PCM DAC". It's not.

Cheers,
SG