[pulseaudio-discuss] Audiophile newbie reporting in

[Maarten Bosmans]

2011/8/3 Michael Cronenworth <mike at cchtml.com>:
> Maarten Bosmans wrote:
>>
>> What would you expect from such an area on the wiki?
>
> It would have answered my questions without having to ask them here. I will
> create a page soon with a summary of what we have discussed.
>
>> Then you can set resample-method in daemon.conf to src-sinc-best-quality.
>
> That's what I figured. Thanks.
>
>> Indeed, Jack will not gain you anything over PA for this scenario.
>
> I guess I should clarify. He wants to use Jack to have bit perfect playback
> of even 44.1k streams. He hates upsampling that much.

If you want to avoid any resampling, just bypass pulseaudio
alltogether and open the alsa soundcard directly. If you do that
right, you have guaranteed bit-perfect playback to the hardware,
without resampling, mixing or software volume adjustments.

>> Probably not beneficial, but I'm not sure, honestly. The only way to
>> know for sure is to listen and compare the difference. I mean, you
>> audiophiles can talk for hours about what's the best way to setup
>> things, but eventually it all comes down to what sounds best.
>
> OK.
>
>> Well do you even understand why 96 kHz is better than 48 kHz? For the
>> same reason 192 is better than 96, be it with diminishing returns.
>> Than reason has nothing to do with our ability to hear ultasonic
>> soundwaves.
>
> Yes, I know the definition of sample rate.

Sure, but why do you want 96 instead of 48 if you can't hear anything
above ~20 kHz?
(warning: trick question)

>> Sure, resampling does something with the audio sampling, but I'm not
>> quite sure what "averages samples" should mean.
>
> Yep. I don't know what it means either. I think he has a personal opinion
> that upsampling is evil.
>
>> Pulse can only have fixed sample rate set for a sink. (There are some
>> patches floating around to fix that, which sound like they would fix
>> your single-source use case perfectly, but they land post 1.0 at the
>> earliest) Given that, 96kHz is obviously a better choice for a fixed
>> sample rate, as you'd rather upsample your CD-quality stuff than
>> downsample the 96 kHz recordings.
>>
>
> Good to hear. Thanks for your comments.

I think Arun said on IRC that pass-through would also work for PCM
audio. (or may I'm completely wrong here)

Maarten