signal-noise test

T

timtimtim

Member
I did a signal-to-noise test by recording 3kHz on a zoom portable recorder and winding down the level until I could no longer hear it above the hiss when replayed in Reaper. When recorded in 16 bits the results were very close to what one would expect from 16 bits, that is to say 96dB. But when I repeated the test recording in 24 bits the results were almost exactly the same, though the noise should have been 48dB better. What is happening here? Will Reaper not replay 24 bit recordings? Or am I wrong in assuming 24 bits is 48dB quieter than 16 bits?
 
This test is limited by the dynamic range and self noise of your DAC and monitor system. What exactly are you trying to prove?
 
I'm using the Zoom recorder to record film sound and I want the sig-noise to be as good as possible to do justice to my low-noise mic and very-low-noise pre-amp. The recorder will record at 16 bits or 24 bits at various sample rates. But as I said the hiss appears to be almost the same at 16 bits and 24 bits. What would be the point of designing a recorder to run at 24 bits if the ADC etc will only give noise as good as 16 bits? I'm checking the noise in Reaper because I trust my computer DAC etc more than I would trust the replay circuitry of the recorder. My computer uses the m-audio Audiophile 2496 soundcard.
 
Last edited:
The limiting factor really is going to be the analog electronics around the conversion. There are any number of reasons that the dynamic range of the analog side would be less than the digital could handle, but for the most part it comes down to cost. Design costs, parts costs, etc...

That said, it's very unlikely that the electronics have worse dynamic range than the actual recording environment. In most cases with live mics and anything short of REALLY LOUD sources, the S/N of the space is going to be much worse than any of the gear.

Reaper itself has an absurd amount of dynamic range, so that's definitely not where you're losing resolution.
 
If you are doing film sound, set the zoom to 24/48, get as close to the source as possible, and try to avoid maxing out the Zoom's gain.
 
ashcat: yes I know that the s/n of the environment (I'm recording actors speaking, not music) is often going to be higher than the noise of the electronics, I just want the recording system to be as good as possible.

gecko: I tried exporting the 24-bit test from Reaper at +12dB gain, and the noise simply gets louder, so this seems to show that the noise exists in the recording rather than in the replay circuitry of my computer. So why would recording at 24/48 be better than 16/44?

Thanks to you both
 
24 bit because why not and 48KHz because it's the standard for film, though I suppose you could check with whoever you'll be delivering it to just to make sure. You want to avoid sample rate conversion is at all possible.
 
OK thanks. Can I just check on a theoretical point: if all the analogue electronics was noise free and the A/D converter was also noise free apart from the inevitable quantizing noise, would 24 bit be 48dB quieter than 16 bit?
 
You must log in or register to reply here.
Back
Top