I have question.

Higher amplitude and frequency means higher bit needed to represent the signal. Is that right?

What you are saying makes no sense whatsoever.

The amplitude is adjusted before the ADC. Your 16 bits signal, for example, can represent a 10 mV, a 1 V, or a 1000 V signal. You don't need more bits to record a louder signal, you need an analog to digital converter whose input is suited for strong signals.

You need a higher sample rate to record higher frequencies. It means higher bitrate, but not higher bits per sample.

Amplitude just needs a bigger amplifier. You can make a 1 bit signal very, very loud with a strong amp

SNR requires more bits though.

Ah, but you never mentioned the encoding format.

In what circumstance wouldn't it be true?

sigma-delta???

Uhm, those just trade SNR in one frequency range for SNR in another one.

But then again, 1 bit/2.8 MHz gives quite huge SNR over any modern 24 bit/96 kHz system. Right?

This leads to my own question. Does Ogg Vorbis support multibit encoding? I mean, if I record something with say Korg MR-1, can I then encode the result with 24 bit/192 kHz Q10 settings? Does Ogg Vorbis handle such files? And would they be playable anywhere?

Actually, it'd be a lot less. 24 bit gives a constant 144dB, while I believe 2.8MHz DSD maxes out at 120dB and decays from there.



Of course, PCM is supported.



Provided whatever software you're using to convert from DSD to 24 bit 192kHz works fine, sure. If you're asking if Vorbis will handle that conversion for you, then no it won't.

So SNR-wise 1-bit/2.8MHz (Korg MR-1) is less than 24-bit/96 kHz (Zoom H4)? That's interesting, I'm just weighing different options for purchasing a real portable digital recorder that would be at least somewhat futureproof.

How about the frequency response then? The Korg (advertisement) dokumentation states that the 1-bit/2.8MHz format is far superior to a 24-bit/96 kHz recording, giving better results than an analog tape.

Knowing me I'm not a dog, at least as far as I know that, would there be any real advantage with the 40 kHz (Zoom H4) vs. 50 Hz-100 kHz (Korg MR-1) frequency response? It starts to sound that 24-bit/96 kHz wouln't be that bad after all.


Ok, that was what I ment.

Unless you're not recording audio, I probably wouldn't care about frequency response, as long as everything supported 20-20KHz. If you're recording scientific signals, having more bandwidth might be useful (though I'd recommend a real DAQ device in that case).

How about the ultrasonic harmonics, that is if they resonate with each other in the audible area? They must be captureable in those frequencies a lot better than in the audible area, musn't they?

All they can really do is create beat frequencies, which is a pretty annoying artifact.

If any higher frequencies have a detectable result in the audible range, that result will be captured by a 44.1kHz recording.

So if I want to make an Ogg Vorbis encoded file that in theory could sound better than a 16bit/44.1kHz Audio CD, the way to do it is to use 24bit/44.1kHz for recording settings instead?

There are physical differences (external world). However when you put in the "sound better than" you are into perception (internal world). It is quite uncertain that the very real physical differences are of any consequence what-so-ever for music.

It is possible to demonstrate hearable differences with simple test tones under controlled conditions, but I have yet to find any music where I can tell a difference. I think I have ask for recommendations and samples on every audio forum I've visited where the subject comes up (it always comes up -- repeatedly). The arguments always suddenly stop when I post such a challenge because none of the proponents of more is better can find an example where that is true.

When considering recording, getting in the air music into digital form, there are a couple of dozen factors that will have a much larger impact on how it sounds than a greater bit depth.

Recording at the greater bit depth can be important because of another definitely real characteristic: quantization errors. These occurs for almost every operation on digital audio data. At 16 bit they can become unpleasantly audible. At 24 bit, or better, 32 bit, they are too small to matter. However, after all mixing and mastering processes are completed, with proper dithering, 24 bit can be resampled to 16 bit without degrading the "sound."