+- MinimServer Forum (https://forum.minimserver.com)
+-- Forum: MinimServer (/forumdisplay.php?fid=1)
+--- Forum: General (/forumdisplay.php?fid=2)
+--- Thread: MinimServer Digital Attenuation (/showthread.php?tid=7401)
MinimServer Digital Attenuation - jorgemg1984 - 12-12-2024 14:05
Hi!
I've been using MinimServer for quite a whilte and recently started using the premium version.
I'm doing some digital attenuation to help with intersample overs and using the "convOut=-af volume=-6dB" code in stream.options, using the transcoding to wav32 and ffmpeg.
I would like to know, if possible, how is this digital attenuation done. I know Roon and Foobar do it at 64-bit floating point, and ESS Sabre chips do it at 32-bit and all these options have no impact on sound quality - in all mi tests it even improves because of the extra digital headroom.
In MinimServer, assuming it's done after transcoding, is it done at 32-bit? Is it lossless? To my ears, I hear no degradation, just curious.
Many thanks!
RE: MinimServer Digital Attenuation - simoncn - 12-12-2024 16:46
MinimServer uses FFmpeg to do attenuation and transcoding, so this depends on the internals of FFmpeg and also on what hardware platform you are using to run MinimServer and FFmpeg.
MinimServer tells FFmpeg to decode the input audio to floating point samples so that attenuation and/or transcoding is done using floating-point arithmetic but I don't know whether this is done with 64-bit or 32-bit precision.
To the best of my knowledge the order of processing should be this:
input stream -> floating point -> attenuation -> transcoding -> output stream
RE: MinimServer Digital Attenuation - jorgemg1984 - 12-12-2024 18:54
Hi, Simon
Many thanks for the reply. Floating point is a good idea, that's reassuring. My current NAS is 32-bit, so it's limited to that - but it's good to know that on a 64-bit architecture it may be (eventually) done 64-bit floating point.
As I said, subjectively I hear no degradation.