Hello,
I saw that the re-sampling keeps the bit deepth, e.g. 16bit for CD-signals. It will up-sample the frequency to the given setting.
Is there any chance to increase frequency and bit depth from 16 bit to 24bit?
Best regards
KJM
Why?
It will only add 8 zeroes at the bottom. It doesn’t add resolution or dynamic range. You can’t stretch the dynamic range. Oversampling can make the edges smoother, depending on what algorithm you use.
16 bit is more than enough for domestic use given that playing music over 100dB will likely get you in trouble with law enforcement. 16b =^ 96dB plus your noise floor you can play 116dB without hearing any noise. That’s like standing near a jackhammer or jet taking off. Or your ears permanently damaged.
1 Like
Hello,
I’m not not sure about the dynamic range. I’m using Anthem AVM 90 as processor with ARC. The room correction reduces the dynamic range of the music. Anthem D2 and Anthem STR amplifiers use upsampling to 192Khz/24Bit to counteract this loss of dynamic range. The AVM 90 has no upsampling function to my knowledge.
Ok that makes the question a lot more interesting. But also complicated. However I don’t think it will matter.
So here’s why. So if the absolute level is 0dB, that means no attenuation on the power amplifier, your dynamic range is 96dB. That’s absolute between the minimal and maximal amplitude. So the x-axis (y=0) and y=max=Pmax of your amp. It can’t drop below y=0 (ie absolute). This has nothing to do with noise floor even though it’s close, somewhere near depending on equipment.
If your room correction is calculating it shifts all frequencies down in volume to the total of the maximum correction. This is compulsory because your trouble region that needs extra volume cannot exceed 0dB. So first everything has to come down. That means you lose dynamic range.
Let’s say you measure in a spot where the bass cancels out. Worst case it’s reversed phase and leaves you 0dB where you want 80dB. That means your room correction want to amplify an extra 80 dB. So you entire curve has to drop that in order to push your bass area up that much. That leaves 96-80=16dB for your music. Now everything has to be amplified an extra 80dB. And your power amplifier has to produce 100,000,000 times the power to maintain your music on the same level. And then your bass will still cancel out because 160dB-160dB is still 0.
Erm, now I’ve proven that electronic room correction doesn’t work. It is much more effective to reposition your speakers, use speakers with a decent crossover or get more than 1 subwoofer.
But to answer your question further, you are correct that room correction needs more than 16 bit. It needs a lower floor so it adds 000000000 to the bottom of where the 16 bit started. It extends the 16bit to 24 bit and then starts calculating. It would be a very dumb design if it would depend on the user to feed it 24 bit.
To make a very simple analogy, it takes your Mona Lisa, pastes it on a much larger canvas and then starts repositioning it as needed and then cuts it out and frames it anew. It would be silly to reposition the picture, cut it out, paste it on a fitting canvas and then reframe. That would lose a lot off two of the edges and leave you with blank space.
Lessons to be learned:
Leave the worrying to engineers
Don’t expect IT geeks and electronics engineers to fix acoustic problems. Like you don’t call an electrician to fix the plumbing.
Resampling is useless opration and can only degrade the final sound. Any operation with the data influences the sound, mostly to worse than before.
1 Like
There are many differing opinions regarding this subject. Even Hifi Rose is offering the re-sampling option. I will ask Anthem.
Thx for your answers.
Best regards
KJM
There are no opinions when it comes to math. Either it’s true or false.
It’s true that upsampling can improve the sound. Some people use Roon to upsample and recode to crazy big dsd256 and they say it sounds better. It is making sense for a certain kind of dac (sigma delta) because 1 you take away its own upsampling algorithm and then 2 make filtering easier. Others (like me) prefer r2r and NOS.
I just bought a pair of magnepan speakers and got a big bump in the bass. Very tiresome. It sounds amazingly clear and open. But strangely I am also missing details I could hear before. I did try equalising and that helped for the bump. But not to the same result as repositioning the speakers.
There is no free ride. Fix problems at the root. Acoustics are a real thing.
Who sad and proved, either sonically or mathematically, upsampling improves the sound? Actually it is the other way around, any change in a basic signal will bring more issues when converting to analog domain thus much worse sound agains the raw one.
It depends on what type of decoding you use. So you are theoretically right but in practice it’s a lot more complicated.
If you decode with a ladder of closely calibrated resistors (ie an R2R DAC, whether discrete or on a chip) there is a limited amount of high frequency noise generated. This is the slope minus the steps. Very simple to filter out with a low pass filter. But… Here comes the caveat… Filters alter phase response and cause a high roll-off. Especially the old brickwall analog filter used in the beginning of cd’s.
Personally, I think they should not have used those at all, with music you won’t even notice it. It does however look bad on the specs used for marketing when using 0dB pilot tones going up to 20kHz.
So I altogether removed all filters in my R2R DAC. And it sounds more natural and open than with filter.
The thing is: tightly calibrated resistors are hard to produce. And in chip production there will be a lot of loss of production that doesn’t pass quality control.
That is why the sigma delta (that is mathematical language for the sum of tiny differences) conversion was introduced. It uses just a few resistors that switch very fast.
This works fine but there are different downsides.
1 It produces a lot of high frequency noise that really has to be filtered out. It needs a lot of filtering.
2 the output voltage is much lower so it needs an output amplification stage. This is not part of the dac-chip but depends on the device engineer, on how well you make that opamp stage.
So with both types of dac the simplest way to improve filtering is by pushing the high frequencies up, away from the audible band. So you can filter with less steep slopes with less detrimental effect on highs and phase integrity. So what they do is multiply the clock frequency and interpolate (calculate) the steps in-between values. This is called oversampling.
This will have no effect on an old R2R DAC with an analog filter. But it makes a ‘modern’ sigma delta dac sound a whole lot better.
I didn’t talk about what comes before the dac itself. A good clear signal with an accurate clock is important too to avoid misinterpretation. All that can be handled while upsampling.
Now this is all very nice with CD quality 16-44 files. But like you said: it doesn’t add any information. That is true. But it does make it easier on the decoding. And that is what you can clearly hear. You can leave oversampling to a chip in the cd-player or dac, but a computer has a lot more computing power. So you can do lots of fancy stuff, just what strikes your fancy.
But if you stop guessing and just record more information, that will make it even easier and sound better too.
That is why I prefer 96kHz PCM for my R2R DAC. Or DSD64 (like on a SACD) which works like sigma delta. Now I don’t know exactly why but DSD recordings sound fenomenal.
To conclude: since 99% of dacs use sigma delta, and Rose Audio uses them, oversampling is useful. If you hook up an external DAC, you can do as you choose. I like to be able to choose for myself. And I hope you’re a bit better equipped now to do the same. 
Well I’m in favor r2r nos with no regular filters :). Using Audio Note DACs have certain filtering in stage transformer and tube stages without influencing the recreated signal. That’s the reason they are most musical and bringing correct timbre and timing in the music against 99% of gear on the market(even on cost no compromise gear)