Jump to content


  • Content Count

  • Joined

  • Last visited

About Miska

  • Rank
    Masters Level Member

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Miska

    Archimago and the TLS DS-1

    Sometimes the content is just that, really boringly produced. And some colorful coloration device makes it sound less boring. However, it doesn't mean it is more resolving, more correct, or more accurate... That's why I find measurements as a good baseline check, that things are at least not wrong in any big way. Then from the well measuring devices one can pick up the ones that also sound good... Similar thing for tweaks too, it is good to check that they don't make things actually perform worse... So in my opinion both measurements and listening are very important. Measurements don't tell it all, neither do listening. My opinion is also that if you can measure differences (and in many cases you can), those are the ones that could be more likely ones accounted for the heard differences, not something that eludes the measurements. Not always though, because some measurement could be missing.
  2. Miska

    HQ Player

    HQPlayer doesn't touch hardware volume control, so if there's a separate hardware volume it may need adjustment. But with any processing you shouldn't be driving HQPlayer at 0 dBFS because you risk triggering the limiter. At least keep eye on the "Limited" counter in HQPlayer main window. It should stay at 0 at all times. If it begins to increment, turn down volume in HQPlayer. As recommended in the manual, volume in HQPlayer shouldn't be turned higher than -3 dBFS.
  3. Miska

    HQ Player

    What kind of DAC do you have? If you have volume control in HQPlayer turned high enough, there may be volume control in the DAC itself too. Checking the hardware volume in ALSA mixer levels could help, it defaults to lowest setting for safety reason.
  4. Strange, I can easily max out gigabit network on Windows with SMB file transfers, Intel NICs and all offloads enabled... If the speed falls below that 100 megabytes per second, it is because of the spinning HDD not being able to keep up constant speed... This is on stock Win 10 Pro which is still even latency optimized for multimedia use. Windows Server versions are I/O bandwidth optimized instead and should perform even better on such test.
  5. IOW, if we look just at numbers. From 100 to 352.8 kHz is about one and half octaves. You would need about 100 dB/oct analog reconstruction filter for perfect reconstruction... That is 17th order filter.
  6. These are measured from analog outputs of the entire DAC, not just the chip, the chip output would have notably more. So far, I have not seen a DAC that would have so much analog filter that it would remove these images completely. Otherwise they will begin to suffer from phase response errors already in audio band. For that, and hires support reasons they need to have -3 dB point around 100 kHz. 8x digital oversampling filter is just doesn't move images high enough. As I have shown earlier, running the digital filters externally and the DAC with 705.6/768 input already reduces images significantly because it is twice higher rate than the on-chip digital filters manage to do. Holo Spring 2 when running R2R PCM section at 1.4 MHz rate has enough filter to remove images that would be starting at that frequency. That's one reason why I use 256x digital oversampling filters and DSD output with ADI-2... No images at all...
  7. Of course you have output filter like with analog class-D too. Higher the frequency, easier it is to remove it with simple filter without effects near audio band...
  8. At least it doesn't harm. Many DACs give lowest THD+N at -10 dBFS. You probably still don't have analog volume control turned up to max. So it is attenuating the signal to be amplified again by the headphone amp stage. You get better quality when you don't try to push the DAC to the max just to get attenuated more at the following volume control stage... So a form of "gain matching".
  9. Merging for example does 8 channels of DSD256 over ethernet, that is 86 Mbps. Stereo DSD1024 is the same 86 Mbps...
  10. Miska

    HQ Player

    No, it is too late to do it at that point...
  11. Yes, it's been supported for a while already, but only recently devices have appeared that also support it. This device is "a little bit outside of my budget", but I've been testing DSD1024 on Holo Spring 2 DAC. It takes about twice as much power as DSD512. With -2s filters and/or with help of recent GPUs it is not too hard to reach though. I've done most of my testing and measurements with poly-sinc-ext2.
  12. It is not that device's biggest problem. I would be more worried about the amount semi-correlated noise it puts out in the MHz range. It needs better analog filter... I bet it uses one of those DAC chips that is advertised to have "full integrated low-pass filter". Old CS4328 was like that and it also puts out peak at 2.8 MHz, but less, but OTOH it wasn't designed for hires, so it cuts much more aggressively right above 20 kHz. Meridian Explorer2 also cleans up little when running at lower level than 0 dBFS. And much more so when receiving content upsampled to 192k (bypassing some of the MQA filters). Overall, I'd say it is bad idea to run any DAC to 0 dBFS. Even if not doing upsampling, dithered volume at -3 dBFS leaves some headroom for the DAC's DSP especially when you run in some digitally clipped content (common for modern RedBook). Some DACs like Benchmark DAC3 enforce this on you at DAC side. For this purpose, top of the line Wolfson DAC chips have configuration option for -2 dBFS digital pre-pad, but IIRC it is disabled by default. But I have some material where even -3 dBFS is not enough...
  13. There are some systems that work like that, Mola-Mola and NAD for example. I don't know exact reasons, maybe because it requires some development effort that spans from DSP domain to electronics, not a small, cheap or easy task. And quite different from traditional designs for fairly conservative market.
  14. This is with Sharp. Since SD Sharp (slow-delay) is minimum-phase version, it gives same frequency response. Same between "slow" and "sd slow". Here's Slow: And NOS: These are with 44.1k input too.
  15. IIRC, there's also a Windows service related to that, is it running? You are using networkaudiod version 3.5.6?