TOPIC: Crazy? Lower volume in SGear = lower resolution?

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13537

Sorry if this is a stupid or crazy question... I just stumbled on this Sound On Sound article... and after reading it realized it was from 1995 (!). This article stated " because Controller 7 turns down a sound digitally, the lower the volume, the lower the resolution of the sound, because fewer bits are used to represent it. As you lower the resolution of a digital signal, the quantisation noise and distortion increases."

I'm wondering if it applies to today's software, in particular S Gear. (If it does, then probably all software is subject to the same effect... I'm not suggesting that S Gear would be the only one affected by any means).

I'm using an FCB 1010 expression pedal to control the main volume in S Gear (either the output of Pro Convolver or the I/O trim output). I happened to choose cc 7, value 65-127, but of course could easily choose another cc.

Is that statement true of S Gear, for my purposes? I think not, since I'm not really using cc 7 to lower the internal volume of the "unit"... I could use any other cc command number, and it's just operating an S Gear control parameter. Most important I sure don't hear any quality loss.

So I'm guessing this is mixing apples and oranges... and/or may be so outdated as to not apply. Mike, can you (or anyone) verify this old info doesn't apply to S Gear (and virtually all of today's audio software)? Thanks!
  • gearhead001
  • gearhead001's Avatar
  • Offline
  • Tone Master
  • Posts: 71
Last edit: 5 years 3 months ago by gearhead001.

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13538

We did grapple with that in '95 when 16 bits was common but with 24 bits we've more or less forgotten about it because it's essentially inaudible. You can record at very low volumes and the noise floor will never be an issue.

Play away! :rock out:

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13539

I didn't read that article, so I could be off a bit on this brief analysis, but...

First this has nothing to do with MIDI CC 7 or any other CC. It's only that it's being used to control the volume in the digital domain that has the impact on sound resolution. You'd get the same result from any control, MIDI or otherwise that controlled the volume in the digital domain (between the input A2D and output D2A converters).

Second, the "resolution of the sound" involves two dimensions - amplitude and frequency. When converting an analog signal to digital, the sampling frequency determines the resolution of the wave form in the frequency domain. The standard CD sampling rate of 44.1kHz establishes the frequency range that the digital domain is capabile of handling, which is 1/2 the sampling rate. This is well above typical human hearing range, and generally not a problem. Higher sampling rates are certainly better, but may be a bit of marketing hype. I record everything at 44.1.

The other dimension is amplitude which is the number of bits available to capture the amplitude difference from the quietest to the loudest signal. In the old days, that was 16 bits which limited the dynamic range. Like any system with limited dynamic range, the delta between the noise floor and the loudest signal is limited. The smaller the delta, the less difference there is between the noise floor and the loudest signal.

With 24 bits, there's ample dynamic range in the digital domain so that noise floor is no longer much of an issue. With 24 bits we can therefore set input levels lower without introducing noise, get better dynamic range from the instruments, and avoid digital clipping - which is never a good thing.
The following user(s) said Thank You: qryss

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13540

Thank you, Jamsden. What a wonderfully succinct explanation.

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13541

The original point - lower volume in the digital domain = lower resolution remains true. In practical terms the impact is reduced these days because 24 bit audio has 48 dB more headroom and decent converters (say 120 dB S/N) will still have 96 dB available above the noise floor if you have recorded 24 dB below digital clippling/have turned the output down 24 dB in the digital domain.

There are limits to this of course: If you are getting up toward 48 dB down (below digital clipping) then you are in the domain of 12 bit audio resolution. 72 dB down is 8 bits of signal and 4 bits of noise on the same 120 dB S/N converters (which are pretty high end - lots of consumer stuff is worse 108 to 115 dB s/n). The graininess should be evident with enough amplification in the analog domain.

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13578

Thanks, that makes sense. Still lovin' SG!
  • gearhead001
  • gearhead001's Avatar
  • Offline
  • Tone Master
  • Posts: 71

Crazy? Lower volume in SGear = lower resolution? 5 years 3 months ago #13599

Many HB guitar pickups have hum at ~85dB down. Some designs are much better. Considering the SNR of pro studio analog recordings of old being ~90dB SNR, 108-115dB SNR converters are exceptional. Some prefer one converter over another, but it’s hard to tell preference from technical transparency. So many factors to consider when comparing equipment with ~120dB+ SNR. Our brains can be “tricked” into preferring something because of the color scheme of a logo or what not.

I think it’s more important to have a good “front end”. My Presonus Firebox has decent low noise FET Hi-Z preamps, but the converters generate harmonics on 1kHz, 2k, 3k…that peak on 8k at -100dB RMS. It does become audible with lots of gain. Even though the 2k is at -105dB, it’s probably more prominent to human ears. I might try relocating the FB away from my PC now that you mention it. Thanks for that...and I’m glad I thought of it ;).

I am now convinced (thanks to elambo) that recording at 88.1~96k is a good idea - - both because the nyquist+ freq filtering is much smoother & higher than at 44.2~48k, and the error rate of the parallel conversion chips is higher at higher SR’s. Newer converters may continue to decrease error rates, making recording at even higher SR’s possible without any compromises, but is there any point when there is virtually no harmonic content up that high? Most certainly not with electric guitars, anyway. I’ve found that since many of my plugins use multi x oversampling which usually tops off at a certain rate (384k?), I only use ~25% more CPU going from 48k to 96k. I’d record at 88.1k to save more CPU, but I’m afraid the odd math calculations may negatively affect real-time conversion, midi sync, or something else - - probably not, but I err on the side of caution on that.
  • GCKelloch
  • GCKelloch's Avatar
  • Offline
  • Tone King
  • Posts: 526
System Specs:
MB: ASRock H470M-HDV/M.2
M2 SSD Card: 256GB
CPU: Intel i3-10100F
RAM: 16GB DDR4-2666
GPU: Nvidia Gforce GT 1030
Last edit: 5 years 3 months ago by GCKelloch.
Time to create page: 0.069 seconds
Powered by Kunena Forum