[rmp_fyf]

Hi all;

(Really enjoying Kodi! Longtime XBMC user).

Having a bit of an issue with hires FLAC files. My CD's are ripped on another Linux machine (using flac through abcde) though I've got some really nice piano recordings bought from 2L.no - some are 96/24 2.0, some are 192/24 5.1.

The computer runs Ubuntu 14.04 and is connected asynchronous USB to a nice DAC. The DAC will accept a number of input frequencies. Nothing is stored on the PC, my content lives on a NAS in another room. The PC is very understressed, usually hovers well under 20% CPU utilisation and over 6GB free RAM.

Was running Pulseaudio with best quality resampling filters (src-sinc-best-quality) though I've since disabled Pulseaudio and am running ALSA direct to hardware (no mixer), which on Redbook content sounds great, and also good on 48kHz sources. Raising the source frequency higher starts to introduce sibilance, it's audibly notable on 96kHz content and awful on 192kHz. This doesen't change when I'm running ALSA direct to hardware or whether I'm running a moderately fast resampler in PulseAudio (e.g. speex-5 or speex-10).

Tearing my hair out with what this might be - my previous install (XBMC pre-Kodi) didn't do this (it also used another DAC, though the testing above would seem to exclude this as a possibility, albeit indirectly).

Could it have something to do with the way FLAC is played/uncompressed/etc in Kodi? Is there anything I can do to change it? (Have I missed some sort of 'audiophile settings'?) I note there's a resample engine in audio settings - does stopping Pulseaudio render this ineffective?

Really keen to get this working properly! The 2L.no tracks sound absolutely amazing for any audiophiles out there (totally worth the purchase). Any help appreciated.

[fritsch]

Provide a Debug Log. When kodi's volume is 100%. We don't touch the flac at all. ffmpeg gedoces it and we push it onto the sink in the highest format available.

"Raising the source frequency"? <- are you sure that you don't introduce them by "upsampling" 44.1 khz content to insane 192 khz? Why would one do that, when ripping CDs?

[noggin]

rmp_fyf - have you played 192kHz content on your current DAC without sibilance in any other setup?

(There are reports that some 192kHz content contains ultrasonic content which can cause harmonics and intermodulation in some set-ups)

Whether you agree with the discussion of 192kHz content or not - this could have some interesting test content in the intermod test section : http://people.xiph.org/~xiphmont/demo/neil-young.html

[rmp_fyf]

Thank you both for responding.

I feel like a bit of an idiot! noggin has it - my new DAC is a vintage Redbook design, and whilst it does oversample it's not at a frequency allowing 96kHz or 192kHz without truncation and other odd things. When my Kodi settings are setup to resample it's no surprise it doesn't sound brilliant, I've been comparing bit perfect with on-the-fly resampling of some pretty serious data in realtime, so yes, it's not going to sound as awesome. Not a surprise.

The giveaway was that 88.2kHz content plays fantastically, 96kHz does not

I'll go away and deal with ALSA insisting to route out at 32-bit depth, and some time later look at MPD. Essentially it'd be epic if I could configure this thing to play Redbook and 88.2kHz audio bit-perfect and resample all else (something Pulseaudio doesn't do).

And yes - those 192kHz files are legit. 2L.no sells them, on an appropriate DAC they sound very good super dynamic range is your thing... if

[noggin]

Thank you both for responding.

I feel like a bit of an idiot! noggin has it - my new DAC is a vintage Redbook design, and whilst it does oversample it's not at a frequency allowing 96kHz or 192kHz without truncation and other odd things. When my Kodi settings are setup to resample it's no surprise it doesn't sound brilliant, I've been comparing bit perfect with on-the-fly resampling of some pretty serious data in realtime, so yes, it's not going to sound as awesome. Not a surprise.

The giveaway was that 88.2kHz content plays fantastically, 96kHz does not

I'll go away and deal with ALSA insisting to route out at 32-bit depth, and some time later look at MPD. Essentially it'd be epic if I could configure this thing to play Redbook and 88.2kHz audio bit-perfect and resample all else (something Pulseaudio doesn't do).

And yes - those 192kHz files are legit. 2L.no sells them, on an appropriate DAC they sound very good super dynamic range is your thing... if

Yep - though presumably it is the bit depth not the sampling rate that improves the dynamic range. Or just that the recordings are less compressed and more 'cared for' than many 48k/16bit recordings.

[rmp_fyf]

Yep - though presumably it is the bit depth not the sampling rate that improves the dynamic range. Or just that the recordings are less compressed and more 'cared for' than many 48k/16bit recordings.

It is, though you'd swear 2L exaggerates it in their piano recordings. Whomever's mastering is very good - very much worth a listen.

[noggin]

Yep - though presumably it is the bit depth not the sampling rate that improves the dynamic range. Or just that the recordings are less compressed and more 'cared for' than many 48k/16bit recordings.

It is, though you'd swear 2L exaggerates it in their piano recordings. Whomever's mastering is very good - very much worth a listen.

I work in the broadcast industry. It's amazing how good 48/16 can sound when someone who knows what they are doing is balancing and not pushing everything through lots of compression to make it SOUND LOUDER THAN ANYTHING ELSE

Heck 32k/14-10bit companded (aka NICAM 728 - the digital sound system we used to use with analogue TV) sounded brilliant when fed a decent mix (often sounded better to me than the MP2 version when I could compare DVB-T with MP2 at 256k 48/16 and PAL-I with NICAM 728k 32/14-10)

I suspect the 2L stuff would sound great at 48/16 too...

[rmp_fyf]

Nice. I'm an engineer with a good backgrounding signal processing... though it's not the mainstay of my work. Audio is a hobby; Kodi/Linux makes life easy and fun to play around with (for those of us old enough to have CD collections - or to still collect CDs! - its tremendous fun).

On the old DAC I'd played with that very idea and decided I couldn't really reliably tell the difference between 96kHz and 192kHz sources. This particular album I'd bought as a bit of an experiment - they sold it, I thought I'd buy it and hear for myself. Not to suggest there was no difference: there indeed was a difference, I just couldn't reliably pick out "better" as the 192kHz version of the same track.

The new DAC is inherently Redbook so there ends that argument of ever-increasing sample frequency. It even has a higher noise floor than the last. It sounds better, however, considerably more 'the singer or instrument is in the room'. "Better" isn't necessarily a function of ultimate detail, and too much compression really is a killjoy.

I guess at this point one tends to value their good recordings very much.

[un1versal]

This is a joke of sorts and a slight trolling moment, so excuse me in advance., but Ill tell it anyway.
I know a guy who thought he use to be GOD, so that trumps all lower life forms and occupations, but apparently lost his omniscient knowledge of the infinite universe and omnipotence over "his" creations. He too like to tinker with all sorts and never lost an opportunity to brag about his former occupation . Long story short, now lives in a nice padded room with his new imaginary friends.
One thing though I think he is an audiophile also.

[rmp_fyf]

Sorry, don't get it. I just came online to sort out a problem only to realise I'd made an obvious blunder, which someone helped point out (for which I'm thankful).

I know very few true audiophiles... they all seem to refuse to use a PC as a source, and claim to be able to pick out audible differences between FLAC and WAV when pressed to use one...

[fritsch]

Can you post a Debug Log never the less? I want to see if the Audio Device (driver wise) gets enumerated correctly, concerning format and available samplerates.