Thanks Speed_Racer, I guess I didn’t really want to get into a technical discussion as the thread is more about what we can hear. I have researched this in the past & my understanding is upsampling is employed for interpolation & more importantly it allows the use of a gentler (less steep) filter hopefully resulting in better sound.
Perhaps this is DAC dependent, some DACS may benefit from this especially older DACS.
Seems Rob Watts design cant be improved upon by HQplayer (to my ears anyway)
Some use HQPlayer with Dave and some sold their MScalar after trying HQPlayer, so its not about your DAC. The reason HQPlayer usually improves the sound is because it has much more advanced filters than is in a DAC because a DAC usually don’t have a very powerful “CPU” in it. For Qutest, once you up-sample to 384khz or more the internal (limited) up-sampling in Qutest will be bypassed.
Try these settings (except volume min/max -80/-14), from what I have read it compares well to MScalar (and yes, its 1M taps just like MScalar):
I call it cleaner sound. Due to the high accuracy apodizing filters. The harshness caused by errors in the source data (due to the processes employed by the recording ADC) are removed and the sound becomes clean. Some may mistake that harshness for “detail”, but it is distortion.
Thanks Jussi,
Its nice to see the developer taking an active interest in our discussions on here.
I have to say I agree somewhat, my initial impressions of HQplayer was that it sounded smoother (more analogue, which is the goal i guess) but on closer listening it sounds like the leading edge of transients are being blurred to achieve this. I’ll keep listening
What does this mean? If you use a linear phase filter the time domain is not altered so I don’t see how the musical information could become smeared.
I have found that using one of Jussi’s apodizing linear phase filters on most source material does nothing but improve clarity while reducing digital harshness. In other words, the detail and clarity is improved but without the hard edge. I prefer Jussi’s poly-sinc-ext2 filter but I am playing with poly-sinc-gauss-long currently. Ny DAC does better with DSD256 input so I upsample all PCM and DSD to DSD256.
After now using the poly-sinc-gauss-xl(a) filters for more than 2 weeks now, I am back to xtr-lp.
Hard to put my finger on but something was bothering me - maybe the space between instruments a little exagarated?
Instead of going back to poly-sinc-xtr-lp , I am now (strated to) try poly-sinc-xtr-lp-short.
Again I am going to give it it some weeks before coming to conclusions.
Maybe I am just burned in too much with poly-sinc-xtr-lp?
I really like the gauss-xl (non apodizing) version, as it sounds balanced and clean in both frequency and time domains. It is kind of similar to closed form which I also enjoy. However, the gauss can indeed sound a bit too distant and spaced sonically like you suggest.
For me, ext3 (or ext2) is the king of realistic timbre and therefore that’s what I use.
After tried gauss-xla for a couple of days for 1x, I went back to ext3. I also felt gauss-xl(a) cause space between instruments a little exaggerated. gauss-xl(a) also does not have the “body” with vocal that the ext3 has.
For Nx, I prefer sinc-xtr-short-mp to ext2
Why? Because many people feel that HQPlayer does a much better job of upsampling than Roon. I know I do. If Roon were to drop HQPlayer support I would drop Roon…and I am a lifetime subscriber!
I had been struggling trying to dial in my Holo May, but the new poly-sinc-gauss-long filter upsampled to 1536kHz with LNS15 just nails it. I am not good at expressing things in terms of long/short, linear/minimum phase, time/frequency domain but this filter seems to have it all - clear, spacious, good separation yet full with solid bass and a nice touch of roundness. Not really any compromises in my very revealing system.