Sony's design for SACD uses the LFE channel differently from everybody else.
Originally Posted by kokishin
I took your advice and converted a stereo dsf file to 24/88 flac using foobar2000. Seems ok.
I recall reading somewhere that there may be some issue converting a multichannel dsf file to flac with respect to the .1 LFE channel. Something to do with the LFE channel requiring +10dB compensation adjustment. Please advise.
Everybody else uses the LFE channel as a place to hold *LOUD* bass. So the content in that channel is recorded -10dB down and has to be boosted +10dB on playback to restore it to the level it should be at compared to the other speaker channels. AVRs, for example, almost universally add +10dB boost to the channel by default.
But SACD (as Sony designed it) does not do that. SACD records LFE without that -10dB attenuation. So if +10dB boost is applied during listening that channel will be too hot.
Now, when playing DSD digital audio content, whatever is processing that can make a guess (almost certainly correct) that this came from an SACD and apply -10dB attenuation to that LFE channel so that a subsequent, default, +10dB boost will leave things "correct".
But when you convert the DSD digital audio to another digital audio format, like LPCM, that provenance is lost. So the AVR, for example can't know not to provide the +10dB boost.
Ideally then, whatever is doing the CONVERSION from DSD to something else should apply -10dB attenuation to the LFE channels AS PART OF that conversion. In your case, that means the tool you are using to produce those FLAC files.
You'll have to ask people familiar with the tool whether it does that by default, or can be set to do that.
This is such an ungodly can of worms that studios have taken countermeasures. Many Studios author their 5.1 SACD content so that there IS NO AUDIO in the .1 channel. I.e, it is really 5.0 authoring packaged as 5.1. *ALL* the bass is authored into the main speaker channels.
Some older SACDs, on the other hand, took the approach that, Sony notwithstanding, they were going to author 5.1 SACD content including the -10dB attenuation of the .1 channel as part of the authoring. I.e., the same thing EVERY OTHER FORMAT was doing with that .1 channel. These are, of course, truly problematic recordings today since the fixes built into the conversion algorithms in modern gear will now attenuate that channel yet ANOTHER -10dB.
Having fun yet?