It is better to adjust this volume from HQPlayer like you’ve done. You can set the maximum volume limit in HQPlayer as well to reduce likelihood of accidents. Please also check that DAC is not set as default audio output device for macOS! For example set the internal speaker as default output for macOS.
1 Like
Thanks! yes SMSL is not the default output for mac I think I saw you mention that before and I fixed it. Glad to know I’m in a safe position with the much low volume on HQPlayer.
Okay, after modifying it to 250ms, it works fine, and therefore no more micro-dropouts.
I also tried it with 100ms, and it works just as well.
Question: I couldn’t find an explanation in the HQPlayer manual on this point, but what is the impact of the buffer choice? Should I try to set it to the minimum that doesn’t cause micro-dropouts, or simply leave it at the maximum of 250ms? Does this have an impact on the sound reproduction, or not?
It is size of the hardware DMA buffer. It also defines how frequently the process needs to wake up to feed more data to the hardware. Lower the value, more transfer overhead there is, since the DMA transfer needs to be programmed every time. With lower values you are also more likely to get drop-outs in audio. Higher the value, more delay there is when you are adjusting volume or seeking, before you hear the change.
100 ms is recommended setting if you need to adjust this. Few exceptions to this though. On Windows with ASIO it should be left to “Default” and the hardware buffer should be adjusted from the ASIO driver control panel instead, in ASIO case “Default” instructs HQPlayer to use the buffer size proposed by the driver (not all drivers even allow changing it). On macOS CoreAudio, this setting is not available at all, macOS manages this by itself.
In addition, there’s the FIFO buffer which is above this hardware buffer. Most of the delay comes from this one since it is usually larger. The “short buffer” setting adjusts size of the FIFO. There you can trade between reliability and delay.
Finally updated on Desktop/Mobile/Tablet version can see “Which HQP Filter are you using” by a simple screenshot! No longer lost in the shadows. (pending to fix DSD sources )
Thank you to all who voted on the request and for Roon’s prompt action!
Edit: Please vote more to request fixing the DSF signal path 
9 Likes
A little hello, news, and food for thoughts for whomever would like to experiment on a rainy weekend (might be me if challenged) ; IMO :
room response has an influence on filter choice ; so has the choice of MP vs LP eQ. I convolve phase linearised FIR filters. No filter right of 816 Hz, none with Q over 5
I upsample every source to SDM 256 with 7EC Super
1x ps xtr short lp (and I listen to a lot of classical, not just the targeted jazz electronics pop rock)
Nx polynomial 1
I believe LP eQ and short filters to be a consistent pairing for liveliness. I’m puzzled by “space” in @jussi_laako 's ratings. I dig and hear that there are more recording environment details with longer filters but I feel instruments placement (soundstage ?) is better with shorter filter but the recommended psg hires (classical on loudspeakers), (often awkward presentation)
Thanks for the quick solution to my problem! I did not do the simple math to realize 40.96Mhz is 1024x DSD. 
Quick solution provided by response below!
I cannot get the AHM7EC8B shaper to work at all. Is there a combination of parameters that is required to make this work? There is no playback at all and it seems my PC is not processing anything when I hit play on Roon. Sometimes HQPlayer will force itself to close when I try to go back in to change settings after I try to play. It seems the best filter and maximum rate I can make work without issue is ASDM7EC-super 512+fs at 48k x512.
I am using Roon – HQPlayer – NAA – Holo Red – USB Holo Spring 3 KTE. I tried Sinc-h and Poly-sinc-gauss-hires-lp as these are my two preferences. DAC Correction is on for the Spring 3. bit rate is 48k x512. PC specs are Nvidia RTX 4090, AMD 5950X, 64GB RAM. Log shows possible issue with output rate? HQPlayer is latest release 5.16.1.
Log Output
& 2025/11/14 12:45:07 Play (-1/0)
2025/11/14 12:45:07 Offload: resampler=disabled convolution=disabled
+ 2025/11/14 12:45:07 Playback engine running
2025/11/14 12:45:07 IntegratorM: FIR2
2025/11/14 12:45:07 No suitable output rate for 96000, stop
2025/11/14 12:45:07 Stop request (tail)
& 2025/11/14 12:45:07 Stop...
& 2025/11/14 12:45:07 ...stopped
2025/11/14 12:45:07 Set volume: -3.010300 +
& 2025/11/14 12:45:08 Play (-1/0)
2025/11/14 12:45:08 Offload: resampler=disabled convolution=disabled
+ 2025/11/14 12:45:08 Playback engine running
2025/11/14 12:45:08 IntegratorM: FIR2
2025/11/14 12:45:08 No suitable output rate for 96000, stop
2025/11/14 12:45:08 Stop request (tail)
& 2025/11/14 12:45:08 Stop...
& 2025/11/14 12:45:08 ...stopped
2025/11/14 12:45:08 Set volume: -3.010300 +
& 2025/11/14 12:48:55 Play (-1/0)
2025/11/14 12:48:55 Offload: resampler=disabled convolution=disabled
+ 2025/11/14 12:48:55 Playback engine running
2025/11/14 12:48:55 IntegratorM: FIR2
2025/11/14 12:48:55 No suitable output rate for 44100, stop
2025/11/14 12:48:55 Stop request (tail)
& 2025/11/14 12:48:55 Stop...
& 2025/11/14 12:48:55 ...stopped
2025/11/14 12:48:55 Set volume: -3.010300 +
& 2025/11/14 12:48:55 Play (-1/0)
2025/11/14 12:48:55 Offload: resampler=disabled convolution=disabled
+ 2025/11/14 12:48:55 Playback engine running
2025/11/14 12:48:55 IntegratorM: FIR2
2025/11/14 12:48:55 No suitable output rate for 44100, stop
2025/11/14 12:48:55 Stop request (tail)
& 2025/11/14 12:48:55 Stop...
& 2025/11/14 12:48:55 ...stopped
2025/11/14 12:48:55 Set volume: -3.010300 +
AHM7EC8B hybrid modulators are designed for DSD1024+ output rates.
1 Like
I wish Roon would fix that HQPlayer signal path display…
I’ve updated new comments for request Here please vote at the top 
of this post to help support the update.
We already got the signal path showing better than before.
Many thanks HQplayer fans!
1 Like
Of all the sources to point out this signal path display error, one would think your input on the matter would be highly credible… but for whatever reason the glitch still lingers…
3 Likes
2 posts were split to a new topic: Is there a way to view my voting summary
I was undecided between two setups I liked on HOLO Spring 3 KTE and correction DAC on :
1- SDM 1024 with AHM7EC8B and the classic filters recommended by Jussi
2- SDM256 with ASDM7EC-fast
Ultimately, I’m going to stick with the SDM256; the SDM1024 seems to smooth out the sound a bit too much.
And I just tested other shapers, and I’m going to stick with the following configuration, which is truly excellent with my system.
2 Likes
At least it’s an easy thing to do, probably less than an hour…
From objective point of view with Spring 3 those two cases are pretty close, with minor weighting differences on different aspects. So certainly best to just subjectively select which one works better.
1 Like
Hi! I don’t know, you can just try, If no more votes,
Just comment “support to fix” to show interest
Laurie Anderson’s “Landfall” with Sinc-L@1024 with ASDM7EC-fast 512+fs with DAC Correction. What a beauty. Yep, we had a chit chat with some CA members, indeed Sinc-L for the strings is unbeatable.
Moving on with the strings, and another awesome recording from Kronos Quartet, this time adding a full flavour with Sinc-L@1024 with ASDM7EC-super 512+fs with DAC Correction
1 Like