Izotope Rx 7 Manual Pdf
Garcia Miller
I've read the manual to understand what each of the controls do but I assume that if I just select the sample rate I want and leave all the controls set at their default values, that should be fine. Would that be correct?
Also, when I master these songs I'll probably be setting my limiter's ceiling at -0.01 with true peak limiting engaged and my true peaks will likely be up at that level. In that case, would it be wise to engage RX's post-limiter?
Lastly, I notice that when I select 44.1 kHz, RX says "no resampling, select a sample rate". It shows this even when the file I have loaded is a 48 kHz file. This seems like an error. If the loaded file was 44.1 kHz, I suppose it would make sense because selecting 44.1 kHz wouldn't result in any resampling. But the file I've loaded is 48 kHz. What's going on here?
Change tag only means it will change the meta data only, i.e. it will appear as if the sample rate has changed, but it has not. This will lead to speed up/down, depending on the playback rate. It's mainly used to fix a tag error.
Use a gentler filter and you'll reduce ringing, but you'll either need to cut a lot earlier (thus reducing high end overall) or you'll get aliasing artifacts as any unfiltered signal above Nyquist folds back into the audible range.
Cutoff shift is a trade-off between reducing the frequency range of the pass-band (what's not filtered) and at the same time attenuating aliasing from the stop-band (what's filtered).
Even though you've already true peak limited your input signal before SRC you could have tiny overs during SRC, so stay vigilant. Do one version without post-limiter first and check for overs if you're not sure.
Even though EBU R128 and ATSC A/85 standards have some leniency in the tolerances I'd be a bit more conservative with the final output level, say -0.2 dBTP when nothing is specified and -1 dBTP (EBU R128) and -2 dBTP (ATSC A/85 and in Japan, last I checked, but they might have aligned with EBU since) when required by the standard.
Filter steepness is the slope of the filter. The steeper and closer to the cutoff point, the closer to the ideal filter frequency range, but you'll also get more resonance, i.e. filter ringing in the time domain.
Use a gentle filter and you'll reduce ringing, but you'll either need to cut a lot earlier (thus reducing high end overall) or you'll get aliasing artifacts as any unfiltered signal above Nyquist folds back into the audible range.
Cutoff shift is a trade-off between attenuating the frequency range of the pass-band (pass-band = what's not filtered) and at the same time attenuating aliasing from the stop-band (what's filtered away). Lower values means less aliasing, but also a smaller frequency range of your actual usable signal, and vice versa.
Pre-ringing lets you go between minimum phase IIR and fully linear phase FIR, where 0 is minimum phase with maximum post filter ringing and no pre-ringing. 1 is fully linear phase with equal distribution of pre and post filter ringing. Anything in-between is so-called intermediate phase.
I can't explain the problem you're experiencing regarding "no resampling" [available], but make sure you're working on the right file or you're in the relevant module/batch window and that the actual audio file you're working on is a different sample rate (read: double check inside RX's file inspector).
I will have to do some research about the "no resampling, select a sample rate" problem. I've tried importing multiple 48 kHz files and I get the same error with all of them when I select 44.1 kHz in RX's SRC. I have emailed Izotope about this.
All that information you've given me is great. However, I feel that if I was to get into nitty gritty of each of those settings, I may overwhelm myself. I'm already a pedantic guy as it is and I have a tendency to spend too much time on things that don't make much of a difference in the end. Therefore, if all I want to do is upsample from 44.1 to 48, do you think it's safe enough for me to just go with RX's default settings for these features instead of getting into the weeds? I'd make an exception for the post-limiter feature as I feel confident enough to know what I'm doing with that.
Noise Shaping - stronger noise shaping is theoretically more transparent - in the sense that it more closely resembles an inverse equal loudness contour (of sorts) - but it can cause higher peaks after dithering. Think of the noise shaping as "hiding" the more audible spectrum of the dither by moving it to less sensitive areas to our ears.
I'm not a big fan of strong noise shaping for anything else than very dynamic, non-limited music, and stick to slightly more "noisy" dithers for most compressed music. The aggressive settings can sound a bit processed or artificial, but this is admittedly a pretty theoretical issue.
TPDF (triangular probability density function) can be obtained by setting noise shaping to none and dither amount to high. This is basically like white noise. Decently effective and no artifacts per se, but a slightly higher noise floor.
Auto-blanking - Turns off dither depending on the circumstances. "When Quantized" is sort of intelligent in the way it detects already dithered signals to avoid double-dipping. "On Silence" (as in literally no input signal) and "Off" should be self-explanatory.
Suppress harmonics - Relevant only if you're not dithering and not noise shaping, and therefore truncating instead (for some reason - not relevant in your case). Enabling this function can lower the harmonic overtones caused by the quantization distortion. I'm not sure when this is relevant as "When Quantized" in auto-blanking would be a better choice when trying to avoid double dithering. Maybe someone else can think of such a scenario.
The different presets are various combinations of the above settings and fairly easy to understand once you understand the above explanations. Now that you do, you can simply make your own combo(s) and ignore the presets.