Mark, hang in there! Many here actually believe that CD is virtually perfect, but there is audible improvement with extended bandwith, etc. We are not all crazy! '-)
At this moment, I am listening to a good CD through the OPPO 105 on my 'big' system for the first time in months. Is sounds OK, BUT it is bettered when I listen to DVD-Audio (24-96) or SACD from the same system. That's what I work at this stuff for.
At this moment, I am listening to a good CD through the OPPO 105 on my 'big' system for the first time in months. Is sounds OK, BUT it is bettered when I listen to DVD-Audio (24-96) or SACD from the same system. That's what I work at this stuff for.
Thanks George for this effort.
Before buzzing off to Bruce Springsteen show setup late this afternoon I had a casual listen from the clothes line/kitchen/nearfield on POS old school one piece stereo powered speaker and found -
- Integer SRC is distinctly preferable to non integer SRC.
- Reduction in SR distinctly reduces audible information.
- Reduction in Bit Depth distinctly reduces audible information.
- Application of dithering distinctly changes audible nature of fundamentals and further introduces audible discordant artifacts.
I can elaborate on the above points as required/asked.
My preference of all the down sampled to 44k examples were the non dithered/non shaped versions including down to 8 bits despite the 'white' quantisation noise/hiss.
To my ear the effective timbre change of noise shaping is unnatural/unacceptable/non musical.
Dan.
Last edited:
For one example,say, starting about 4 minutes in, there are some male vocals that kind of sound like American Indian chanting. In the 24/96 version they are very detailed while remaining smooth sounding at HF, and the vocalist's voice is pretty natural sounding. In the 16/44 undithered version, the vocal is grainy and rough-edged at HF, and even distorted sounding at louder points.
All this talk about windows routing reminds me of the one thing I do not like about my cheap nasty DAC that Mark keeps making a dig at 
. It doesn't have a 24 bit or sample rate light. I have to rely on Alsa which hopefully, like most things linux does what it is told.
But main rig out of commission for upgrades at the moment
. It doesn't have a 24 bit or sample rate light. I have to rely on Alsa which hopefully, like most things linux does what it is told. But main rig out of commission for upgrades at the moment
Its pretty easy to see what Alsa is doing. alsa.opensrc.org
ls /proc/asound/pcm??/sub?/status or something like this. its been a few years since I last had to do this. Alsa is far more user accessible than Windows.
You may also get it from alsamixer. If Alsa is configured correctly and your player connects directly it will pass the PCM directly to the DAC without any changes.
What player are you using?
ls /proc/asound/pcm??/sub?/status or something like this. its been a few years since I last had to do this. Alsa is far more user accessible than Windows.
You may also get it from alsamixer. If Alsa is configured correctly and your player connects directly it will pass the PCM directly to the DAC without any changes.
What player are you using?
MPD. When I interrogate ALSA is says it's pumping out 24/96 so I have to trust it. Now as only 0.1% of my music is 96kHz that's possible not optimal for quality but i set it when I wanted to listen to an HD download and left it there. Was a bit of faff to get it working, but now it just gets on with the job.
The server was the best money I have spent on musical enjoyment in a very long time
Where did you get that idea?
Half band filters are normal FIR filters with the added constraint that the corner
must happen around fs / 4 of the incoming data. That results in symmetrical
coefficients which halves the number of multipliers, and every other coefficient
is 0, which halves the effort again.
The attenuation achievable is still determined by the number and values of the
remaining filter coefficients.
You cannot extend the useable pass/stop band of any digital filter to
exactly Nyquist without getting problems.
Last edited:
I think it was covered as part of my uni course. But of course I may have misunderstood.
That is my understanding yes. And added to that, the corner is at -6dB amplitude. So we have a filter which is only 6dB down at 22.05kHz. So its fs / 4 for the outgoing data, not the incoming data.
This is true, for the stopband. But my comment was about the transition band.
Yet that is precisely what's going on in a half-band filter. Depending on whether you consider the -6dB point to be pass-band or stop-band. I consider it transition band.
Last edited:
Here is what worked for the players we shipped. This was driving SPDIF from a Juli@ card.
Code:
audio_output {
type "alsa"
name "Digital"
device "plughw:0,1" # optional
}You need to adjust this for your hardware settings.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.