Now Sunlight allows DSD256 support, do you prefer DSD256 upsampling over DSD128 ?
It is soon to conclude right now, but I feel a better sound quality in DSD128 on my end. Switching to DSD256 brings too much clarity, finally reverting to DSD128 sounds more natural.
In both cases I’m in ASDM7EC/poly-syn-xtr-mp-2s mode. Just something to note, is that in DSD256 I’ve got some stuttering issues which may be didn’t help to get a positive feedback.
I’ll try other settings but for the moment can’t disconnect to the listening session with current settings; Ted did some miracle again 
I find DSD256 to be superior on my system. The clarity and smoothness I am getting just sounds more “real”. I am using ASDM7EC and poly-sinc-ext2.
2 Likes
Like speed-racer I’d recommend at least trying the non-mp (minimum phase) filters. Minimum phase messes with the phase and hence transients. Whatever you prefer is fine, but many like the non-mp filters.
2 Likes
Thank you both. I’ll try that.
When I used the xtr filters, I did prefer the linear phase versions.
Also, even if you think the DSD256 upsampling has “too much” clarity, give it some time and I think you will soon like it more than the DSD128 upsampling.
Hi! So after several comparaisons on Sunlight at DSD 256 bitrate, I ended to the following
Shaper: DSD7 256+fs
Filter: poly-sync-ext2 (in battle against poly-sync-xtr-lp-2s)
Ext2 seems to bring more control somewhere. And definitively Linear filters sounds better, highs are more elegant and refined (not especialy detailed, but they end in a better way).
The Shaper choice (DSD7) is a compromise while I can not afford to keep the ASDM7EC shaper as my server starts to cut the stream sometimes, not a reliable setting for listening. But I’ve always prefered ASDM7EC in DSD128 bitrate.
This evening (more quiet time) I’ll try a comparaison of my best setting in DSD128 versus this setting in DSD256.
I’m wondering what kind of shaper @tedsmith is using in the FPGA and if it does not worth it to imagine a setting where HQPlayer upsample PCM any rate into PCM 352 and let Ted’s algorithms to do its own magic for DSD conversion ?
Besides that, Sunlight is for me the better DirectstreamSr upgrade I’d ever done, I really like it, we get even better tones/timbres than before. The new clarity delivered sounds natural to my ears after speakers position adjustment. Setting the level at 92 is also a good trick to get a bit smoother and enjoyable sound.
1 Like
So, I could live in DSD128 with ASDM7EC shaper and Ext2 filter, very very good. However I’m feeling a bit better in DSD256 with DSD256+fs (not capable of 7EC) with Ext2 filter. More air.
May be I’ll try to stay in PCM and rely on the DirectStream for internal DSD conversion. From now I prefer enjoy listening music as ever. Once again, this upgrade is pure hapyness! Thanks 
1 Like
Try ASDM5EC. The EC modulators are something special. Also try the 512+fs version. These are less CPU intensive but have most of the magic.
I don’t have a DS any longer but enjoy what my CPU can do with the ‘lesser’ EC modulators into R-2R dac.
Try poly-since-lp too if you find ext2 a bit ‘too much’ on some content.
2 Likes
Hi bret66, thank you for the clues. I agree that EC have something special, I would say “more body” or something like that. I tried the ASDM5EC in DSD Quad rate however I still prefer DSD7th order.
I heard (may be Jussi Laako) that R-2R dac would get benefit of HQPlayer probably with other settings than other DACs, and more generally each DAC implementation would be more prone to be used with some modulator shaper and filter in a different manner.
I’ll try the simple poly-sync-lp (2s), I remenmber I liked at a time the poly-sync-long-lp-2s (edit: the short-lp-2s sounded right for me, not long)
I don’t know if Ted communicates a lot on the filters and shapers he use in the Directstream, surely the best compromise between the power offered by the FPGA and all processing quality he can implement in a cohesive manner along the whole process…
Ted have you thought about implementing some selectable filters in the DS2, Is this practical and would it provide some the same type of customization as HD Player?
There’s only so much memory inside the FPGA - that limits the total number of filter coefficients available, ether 1 huge filter, 2 filters of half that size, … The DS is up against the limits and I’d like the Mk II and TSS to have better filters. There’s still some left and, with enough weird work we could load the filter coefficients separate from the rest of the FPGA code. That’s not high on my list.
On the other hand I’m not going to add, say, minimum phase filters. They mess with the phase of the signal and therefor muffle transients. I’ll also not add filters that alias so they have to have a cutoff below 22.05kHz.
There’s the possibility of some separate filters for color instead of messing with the main upsampling filters. Still these kind of filters can be done in analog after the DAC or a digital processing before the DAC (JRiver, foobar2000, and I believe HQPlayer, etc. all allow for some eqing. I don’t hear people talking about using that feature much.)
3 Likes
A good point, Ted.
There are many players that provide EQ. I do not recall any discussions as to the best target curves or other use of a player’s EQ.
I want my DAC to be absolutely neutral and doing its best to provide exactly what is on the optical disk or file.
3 Likes
Thanks sounds like doing externally is the way to go if that is what a person wants
I’ve used convolution within HQPe for several years. My room sounds oh so flat and boring without these ‘corrections’ or EQ. Of course the filters are near infinitely tunable for preferences or equipment changes. The room and speakers are the very most important though.
On another note, this is an interesting project. PGGB - Offline remastering
That looks interesting are you working with the PGGB software?
No, I haven’t tried it yet. I have no affiliation.
Did someone tried the new poly-sync-ext3 filter from the latest version of HQPlayer ?