Question about KHz

[mshilarious]

ADCs don't rely only on analog lowpass filters. Typically, they have a true sample rate of 6mHz or so, and will have a first-order analog lowpass filter at maybe 1/20th that frequency. To generate the data rate they will use DSP brickwall filters, just as SRCs will do. So it's really an issue of DSP resources, given that an ADC must output with low latency that puts a constraint on a chip that an offline process does not have.

 

[PhilGood]

I'm going to find the Nyquist theory paper. This was discussed before several years ago and everything you guys brought up was refuted.

I'll get back to you.

 

[mshilarious]

I'm going to find the Nyquist theory paper. This was discussed before several years ago and everything you guys brought up was refuted.

I'll get back to you.

Everything I brought up was refuted? Did you actually do a loop test with your own converters? Or just your best SRC? Or you ever code a compressor VST?

Oh, and once you find Lavry's paper, you'll see that he documents the *exact* attenuation to which I referred . . .

http://www.lavryengineering.com/documents/Sampling_Theory.pdf

And he has long argued that 60kHz is a sufficient data rate, which can be demonstrated experimentally. But 44.1kHz does have HF attenuation--the amount depends on implementation, generally older converters are worse, probably because they lacked sufficient DSP to do a really good filter, so they would reduce the number of poles to limit computational complexity. I've done that in my VSTs myself, but they were "color" VSTs, so I got away with it

Anyway, Lavry argues against 192kHz, not 88.2kHz. 44.1kHz is considered "good enough" because nobody really cares about -0.3dB at 20kHz, but as a mathematical fact it's not perfect within a 20kHz passband. As a practical matter, a number of A/D/A or SRCs at 44.1kHz will accumulate HF degradation. Try it and see, and then make sure your production process avoids that fate.

Note that the some of the popular converters of the '90s were more like -2dB at 20kHz . . . get yourself a Behri ADA8000 and see for yourself!


PS for pitch detection I would look for the lowest large peak, and then study its third harmonic to get better accuracy in less time. That would work for most if not all real-world instruments . . . but you can't simply look for the largest peak because a lot of instruments have more prominent first harmonics than fundamentals . . . which would still work for tuning even if the absolute pitch detected was off an octave.

 

[dgatwood]

PS for pitch detection I would look for the lowest large peak, and then study its third harmonic to get better accuracy in less time. That would work for most if not all real-world instruments . . . but you can't simply look for the largest peak because a lot of instruments have more prominent first harmonics than fundamentals . . . which would still work for tuning even if the absolute pitch detected was off an octave.

To be pedantic, you can't *always* look for the lowest peak, either, because it could be that 60 Hz hum from the air conditioner. But point taken.

 

[mshilarious]

To be pedantic, you can't *always* look for the lowest peak, either, because it could be that 60 Hz hum from the air conditioner. But point taken.

That's why I said lowest *large* peak. But you could code to discriminate against 50/60Hz. Or use a balanced guitar so the problem goes away

 

[soundchaser59]

The required steepness of the anti-aliasing and anti-imaging filters will cause some passband attenuation using a 44.1kHz sample rate. This is not snake oil as it is easily verifiable with D/A/D roundtrip of a test signal. It is mostly ameliorated at 48kHz but that depends on the quality of the filter implementation. Some converters (mostly older ones) perform worse than others. It is completely gone at 20kHz with any reasonably modern converter at an 88.1kHz sample rate.

Another consideration is the need for oversampling in processing. Arguably it is better to record at a higher sample rate and use processing that thus does not use oversampling (or uses a minimal amount) rather than having every plug in a chain running its own upsampling and downsampling routines. Beyond the obvious waste of CPU cycles, the effect will be similar to multiple D/A/D roundtrip in terms of passband attenuation. The undesirable alternative is aliasing, which is unfortunately common in several popular processing applications . . .

But then does dithering kill all of that anyway when the tune is mastered and put on a 16 bit cd?

 

[soundchaser59]

Hello ..
I am going to buy a condernsire with 30Hz-20kHz frequency response
And a usb mic port pro with 44.1kHz and 48.0kHz

Does it match together ..
Does it make different if the 2 objects .. one have high KHs and the other have low one

and thx

Or maybe the trolling techniques are getting more sophisticated......