preproman
Headphoneus Supremus
- Joined
- Mar 16, 2011
- Posts
- 10,625
- Likes
- 774
.
Keep in mind that while the roll-off may be -3dB or so at 20KHz, it'll be between -0.5-0.75dB at 10KHz and about -1.5dB at 15KHz. Given most music, the performance of most headphones, and the average hearing ability, this isn't much to worry about.
I spent a lot of time playing around with upsampling methods when I had my Metrum DACs on hand. HQPlayer is fun to mess with, because it lets you try a bunch of different upsampling filters. Some are linear phase, some are minimum phase, and they all have varying amounts of pre and/or post ringing. I think it has a couple filters that aren't supposed to have filter ringing. If you're using a Mac, I think there's a known good audio player with different upsampling filters available. Common upsamplers seem to use linear phase, IME. The more you play around and test various filters, the more you might be able to hear the sound they impart on the DAC (and the Metrums acted as a good "template" to test how these filters sound, in a sense). Yes, you'll likely get a better sense of clarity and detail with upsampling, but whether or not you'll prefer that is up to you. I decided I just preferred NOS on everything, even 16/44.1.
It's expensive and ugly as sin, but XXHighEnd supposedly has a good upsampler designed for NOS DACs (the Phasure DAC, in particular). I believe it avoids filter ringing.
Upsampling will bring some other measurable benefits to the table outside of the frequency response (much like playing higher bit-rate and sampling rate files will improve NOS performance to an extent), but, again, it really comes down to what your ears hear as most pleasing.
Jimi would probably be disappointed to hear this:
Jimi Hendrix' Purple Haze (just try and find a good recording of this), sound much cleaner, with excellent separation -
For those who are interested, I have just post a full review of the Metrum Octave MKII here: http://www.tweak-fi.com/apps/blog/show/42726063-building-a-reference-system-part-1-metrum-octave-mkii-nos-dac
When using the Octave MKII in combination with the Audiophilleo 2, Glenn's OTL amplifier (with a 6SN7 driver and six-6BL7s as output tubes) and the HD800s, the sound is absolutely fantastic.
I felt, nonetheless, that after a (very) long term period of listening that something was lost in the subjective realm.
Those who think that upsampling cannot have any positive impact on sound and can only degrade data should consider this: most, if not all, sigma-delta based DACs already have oversampling filters which are doing … integer upsampling (8x, 16x… ). So there is already upsampling (called oversampling) and digital filtering going on the DACs whether we like or not. Also, one has to keep in mind that the digital filters that are most difficult to construct are those made for the 44.1K (in comparison with 88.2K and 96K+ frequencies) CD/ Redbook format. Programmers have to arbitrate between frequency domain and time domain performance; that is why we see different types of filters: slow roll-off, fast roll-off, minimum phase, intermediate phase…
As a result, it is not straightforward and easy to construct very good sounding digital filters. Some companies, such as Ayre or Meridian to name a few, have developed some interesting sophisticated digital filters and their products are acclaimed by many audio critics. However, those DACs are relatively expensive and are still based on sigma-delta with built-in oversampling.
I don't want to stick my neck out too very far, as I've only had the Octave MkII for a few days, but... while I agree it does a better job with 96/24 than with 44/16, I'm utterly convinced that the Octave MkII does a better job with 44/16 than any OS DAC I've ever owned (including several ESS9023 and one ESS9018 implementation (OPPO HA-1), as well as two CEntrance DACs (Dacmini CX and DACport LX).
I'm concluding that you have to spend a lot of money (> $2k ?) to get an OS DAC that can properly deal with 44.1 khz files.
[snip]
Upsampling:
After a lot of experimentation, I ended up settling on SoX upsampling (on Foobar) with the following parameters:
Upsampling to: 176.4K (for 44.1K files only)
Quality: Best
Passband: 90%
Allow aliasing: checked
Phase: 25% (i.e. intermediate phase)
Given the right conditions (i.e. proper source upstream, as well as transparent amplifiers and headphones downstream), the Octave MKII’s depiction of timber of instruments and voices is absolutely stunning.
[snip]
Then there's the overall naturalness of the Octave MkII - with everything I send through it. My ESSxxxx and CEntrance DACS (and perhaps all affordable OS DACS?) sound absolutely sterile in comparison. But it's not as if the Octave MkII is thick and syrupy. I'm talking about timbre, mostly - with instruments and voices just sounding a lot more realistic than with oversampling DACS.
[snip]
However, what struck me when listening to the Octave MKII was not its ability to portray a big soundstage but, rather, its ability to transport me to the recording venue. On many recordings, you can get a pretty good idea of the size of the recording room not by trying to calculate the size of the room that is being painted in front of you but by measuring the size of the room that you are actually occupying.
But on its own, and fed from the Aqvox powered Audiophilleo 2, the Octave MKII gratified me with the best soundstage I have heard from a digital source.
Actually, the imaging specificity is better than what you get in real life. Or to put it in other words, I have never been in a live event that approached the 3D performance and image specificity of the Octave MKII. There are maybe a few possible explanations. Either we do not listen the same way at home and at various concerts. Another explanation is that using multiple microphones has an impact on the 3D perception, but it has the same surprising and counterintuitive result.
However the Octave MKII is the most transparent DAC I have listened to in my system (or elsewhere).
On a Chesky classical music recording, I was able to hear the traffic noise outside the recording venue.
I really felt some empathy for your statement regarding your experience with upgrading from the Audio-gd DAC19MK3 to the DAC19DSP:
I hate it when observations like this sneak up on me long after upgrading. More often than not, I'm initially thrilled with a change in my system, then get a more mature perspective within a a day or two, but sometimes, it's not until I've had a new piece of gear through many hours of playing a lot of very familiar songs that the truth percolates to the top of my consciousness. Going back to the older component after even a week of "adaptation" is a good technique, however, for ferreting out that which can otherwise go unnoticed.
Sadly, it has taken me years to overcome that arrogance which foolishly assumes, "I can A/B two components and straight away decide which is better."
Yes! I so believe this (bolded parts, above)! The bulk of our libraries, like it or not, are comprised of 44.1kHz files, but you have to spend a lot of money on an oversampling DAC (i.e. sigma-delta) to get filters that can decode 44.1k files as well as a NOS DAC can (i.e. Octave MkII).
Wow, you really dug deeply into upsampling:
You've got me curious now, despite my having previously decided, on Cees Ruijtenberg's cautions, to avoid upsampling. Your obvious awareness and avoidance of pre- and post-ringing has me wanting to try SoX in Foobar. Thanks for sharing your no doubt hard-won parameters.
Again we concur:
It's like eating fresh vegetables vs. frozen!
Another amen - seriously. The Octave MkII takes me to Realityville.
With a good deal of variability, depending on the quality of the recordings, the Octave MkII leaves me feeling I'm no longer listening to recordings - it's more as if I am actually at the original performances. And you certainly don't need binaural recordings or purely acoustic recordings to experience it. That sense of being there can happen even with complex studio mixes, like Michael Jackson's Billie Jean, where the remarkable presence of the backup singers, violins, and trumpet accents have given me goose bumps on occasion. I think this benefit is a byproduct of not processing the signal as much as so many other DACs do.
Whoa! I've not been around the block as much as you have, so I'm thrilled to read this, especially given how much I agree with your other observations.
You are so perceptive! I wish I could write about what I hear, as well as you do. Your thoughts, quoted above, have never crossed my mind, but on reading this, I think you're spot on.
Again, I'm thrilled to read this, given your (obviously) greater experience. I can add that the transparency of the Octave MkII improves greatly across the first week of leaving it turned on, 24/7. Cees Ruijtenberg advised me to leave it on for a solid two weeks (with no need to run data through it.) I'm currently two days away from completing that two-week burn in, but the most dramatic changes happened in the first week - nowhere near as big a change as how a 40-year old tube can rapidly improve, but it has been perhaps the most satisfying burn-in I've ever experienced with solid state gear.
For the life of me, I can't remember which track that is, but I think I know exactly what you're referencing. The one I'm thinking of is the sound of a truck downshifting, coming toward the studio as it moves from far right, behind the performers, to pass by the studio on the right (outside the studio walls, of course), and then continue on its way (to a vanishing point behind your right shoulder).
Thanks again for reinforcing so much of what I'd already concluded on my own, with such a well-written piece.
I'm looking forward to your follow-up articles. I hope I don't miss them...
Mike
OK - got to say - Octave (Mk1) plus Aurix plus T1 = sonic nirvana. Full, present, musical, detailed, deep/wide stage , analogue (as opposed to digital), dynamic - wow this combo has it all! It is the best head fi I have ever heard.