Freesound Forums

Just got a Zoom M4, and wondering about the necessity of low-cut filters.

Prior to 32-bit float, something like an H1 or H1n, especially with auto-gain, could benefit from a low-cut filter, because undesired low-frequency noise could interfere with proper record levels, and it could be hard to remove from a 16 or 24 bit recording.

But with 32-bit float, I'm wondering if it a low-cut filter when recording is just a convenience to streamline the post-production work-flow. Aside from (possibly) some extra steps in post-production, would it most always be better to leave the low-cut filter turned off and “fix it in post”?

I'm just trying to understand if/how there's any technical or quality advantage to recording 32-bit float with a low-cut filter, and I can't think of any benefit, other than streamlining the post-production work-flow.

Interesting question !
Well, in my opinion, it might be better to let the low-cut filter turned of, as you don’t have to worry about clipping your recording, except in those extreme situations in which you’ll overload the preamps of your recorder.
Anyway, you could always adjust the EQ and other parameters in post.

Kevin

On the types of devices we're talking about, I think the the preamps are most/always earlier in the signal chain than a low-cut filter - I think they have to be - so a low-cut filter can't protect the preamps from being overloaded by low-frequencies.

On modern devices, are the low-cut filters earlier in the signal chain than A/D conversion? Or is low-cut filtering done after A/D conversion? eg, the M4 has three different cut-off frequency options for low-cut; I don't think they're implemented in the analog domain, for 4x channels. The cost/complexity/consistency would be prohibitive, compared to a digital low-cut.

Even if low-cut filtering was analog, the circuitry would most certainly require a somewhat normalised post preamp signal level. Low-cut filtering prior to preamps would be tricky and problematic, for a number of reasons.

This becomes another argument in support of not using low-cut when recording with these types of devices. Anything a built-in low-cut filter can do, we can do better in post-production. Still, the only argument I can think of for recording with low-cut would be to streamline the post-production workflow.

Now I'm wondering… With >1,500db of dynamic range, why do 32-bit float recorders even need preamps? Why not just connect the mic directly to a 32-bit float A/D converter?

In my understanding, there shouldn't even be a need for mic/line switch. 70db (or more!) is bupkis here. We can adjust for mic/line levels in post. Give me a “raw” input to the A/D!

Well, not sure... But I guess the problem might come from the A/D converter circuit itself.
In fact, the theoretical 1500+dB range of a 32-bits float recording is way lower in the real life, especially because an A/D converter circuit has a limited dynamic range, which would probably not match the very low signal coming directly from the mic.
But I have to say that I’m a hobbyist, and this is just my own guess. It would be very interesting to get the opinion from a professional.

Cheer !

Kevin

Where does such an interesting number of 1500 dB come from? In my opinion, the theoretical depth is 192 dB. Or am I missing something? In the 32-bit floating point format, there are several parallel analog channels and ADCs, due to this it is possible to create such a depth of dynamic range. At the output, these streams are combined into one, with the appropriate quantization depth. I don’t want to seem rude, but it seems to me that in this case (in 32-bit format), discussing whether or not to filter with a low-cut filter when recording is, to put it mildly, an unnecessary topic. I record on my cheap DR-05x in 16/48 format and never use the built-in filtering. In this case, problems almost never happen, even with difficult recordings. Well, it’s very rare that something happens that cannot be corrected in post-processing (clipping that cannot be corrected). I'm not saying that a filter is completely unnecessary when recording 16/48, as in my case. Regarding saving time during post-processing, I want to share my experience. To save time when post-processing in RX 7, I use external plugins that I connect through the Blue Cat patcher. This allows you to configure a whole chain of plugins at once if necessary, and do everything in one pass. Also in this case, it is valuable that you can save presets “for yourself” in the patcher and instantly load them. I have several presets: for working with spectrum and dynamics, a separate preset for stereo processing, and others. In each case, I make adjustments for the processing of this or that record. All this is more technical than creative processing. Well, with this patcher, you can do parallel processing, with varying degrees of mixing dry/wet signals.

@newlocknew : Here is an interesting article explaining (probably better than me) these theoretical 1500+DB of dynamic range in a 32-bit float file.
https://www.sounddevices.com/32-bit-float-files-explained/
But as you can read, the real dynamic range of such a file is finally limited to 144dB.

In addition, it can be even more limited by the dynamic range of the A/D circuit, especially if it uses only one converter. (

Cheers !

Kevin

Thanks Kevin! I'll take a look.

I only use a DC filter sometimes when I use a waveshaper or something else that makes DC offset. But sometimes I don't even do that because it sounds warmer with the audio clipping from the low frequency sounds