Hiyas, I'm curious if there's a proper method to set highest output gain for decoding DSD sourced material resulting in highest dynamic range without evoking clipping. I'm currently using the RG method (scan per album with zero output gain and floor down to nearest positive value), however not sure if it's really clipping-proof. Thus I'd ask how the "safe" value can be obtained a more reliable way (if the RG method already doesn't bring it).
(https://i.ibb.co/BsMKSCP/foo-input-sacd.png)
If you're outputting DSD, you can't even touch the signal without converting it to PCM and back again.
I'm not sure I understand your question and I don't know what software you're showing in your screenshot...
Doing output gain for SACD / DSD the right way
If you want to "maximize" a file that's called
"normalization" which adjusts the volume up or down as necessary for 0dB peaks.* Obviously, the whole file has to be scanned so normalization can't be done in real-time during playback.
if there's a proper method to set highest output gain for decoding DSD sourced material resulting in highest dynamic range without evoking clipping.
Decoding shouldn't
cause clipping. I assume DSD can go over 0dB (without clipping) but you shouldn't find that in a proper production. If it does go over 0dB you can decode to floating-point WAV/PCM (which can also go over 0dB} and then normalize before converting to regular (integer) PCM so you don't clip your file or DAC.
When you change the volume you don't change the dynamic range unless you clip.** Clipping is a (bad) kind of dynamic compression.
I'm currently using the RG method (scan per album with zero output gain and floor ydown to nearest positive value), however not sure if it's really clipping-pc roof.
"RG" is ReplayGain? ReplayGain pre-scans your file so the peaks and loudness are known before the ReplayGain is applied. Every ReplayGain implementation I've seen has an option to allow or prevent clipping and if you don't allow clipping it won't push the peaks over 0dB and it won't cause clipping.
* Most normalization effects allow you to choose a peak other than 0dB but usually normalization means setting the peaks to100%.
** When you reduce the volume digitally, you are using fewer bits so "mathematically" you are reducing the dynamic range but with any reasonable volume change this is insignificant and not an audible effect, The quantization noise is normally inaudible and lower than the (digitized) analog noise.
This is about converting DSD to PCM
To the best of my knowledge, DSD is 6 db down compared with PCM.
This to avoid overloading of the converter.
You might decide to use the level with this value to "match" PCM.
However, if a recording is really hot, you are glad to have this 6 dB headroom.
I do think using 0 dB is a good choice.
If you want to normalize, let RG or R128 do its job.
The software he's showing is the config preferences panel for the foo_input_sacd plugin of foobar2000. These settings apply when .eg., an SACD .iso file is being converted to PCM
file(s) using f2ks format converter.
I'm not sure I understand his use of Replaygain to set amplification of the conversion, but the way I do it is to open the Console during conversion. I start at +6. Clipping shows up as error messages in the Console. If there's an error I rerun the conversion at +5, repeating as needed until conversion completes with no clipping errors.
One could avoid this and simply convert at +0 dB and use RG, but I can't, because my playlists include raw DTS and AC3 files, whose bitstreams are corrupted when RG is on.
I start at +6. Clipping shows up as error messages in the Console. If there's an error I rerun the conversion at +5, repeating as needed until conversion completes with no clipping errors.
I do the conversion at +0, then find the highest peak in the resulting files and just apply the appropriate gain to those files (they are 24bit after all).
Then I convert to 16/44 :-)