Divider adds it’s own error, so it is not that simple. In addition contribution of a single sample to the signal is higher, lower the sampling rate.
This problem is particularly emphasized by PCM, since you have very few samples of high accuracy, also timing accuracy of single sample needs to be very high.
Of course there are the products from PS Audio. Currently I’m using a Pink Faun Audio Streamer using a I2S connection over HDMI to a Pink Faun DAC 4.32. Really great stuff! By the way: this DAC is a NOS design as I mentioned earlier.
Hi @brian,
When doing PCM up sampling say for example up to 352.8kHz from a 44.1kHz source, does the software goes to the usual 4 tapped pole filters? e.g, one at 44.1k, 88.2k, 176.4k and last one is 352.8k. If yes, what type of digital filters you are using? This is something like most over-sampling digital filters prior to the final DAC stage.
This does matter to me as I use NOS mode all the time for Holo Spring DAC. Thank you.
Our SRC is a polyphase FIR interpolator. 44.1kHz->352.8kHz is accomplished in a single filtering/upsampling operation–so no, it does not go through a cascade of four filters.
Generally, I associate the cascaded approach that you described with more resource-constrained environments, since it allows the use of a longer filter for the more critical early phases when the data rate is lower, and then fall back to much shorter (and thus less expensive) filter for the later phases when the data rate is higher, but the filtering requirements are less stringent. This is cheaper than achieving the same frequency response it in one filtering pass, despite the multi-step process involved.
What with thunderbolt DACs? Its alternative, not rare like ethernet/i2s via hdmi products. We have now thunderbolt 3 standard in every modern, powerfull PC. So?
You wrote about fibre, thunderbolt is smth like that (point to point solution).
I used Acourate to create the Roon convolution filter, not trivial but possible to figure out even without any expertise. I applied it to a 15’ by 15’ room with bookshelf speakers, the results were not magic but definitely worthwhile. An amplitude measurement doesn’t tell everything, but these before-and-after curves reflect the subjective effect in that room (the tilt is intentional):
Nice curves, bet it sounds a lot better.
Once you get a response like that, its probably as good as its going to get.
Is this all DSP or did you need to do room treatments as well?
Yes, it sounds nice. No room treatment, just DSP. The speakers were already good, this was about the room. The whole procedure took 20 minutes, including finding the mic and setting it up.
I had already done my main room, which is where the learnings were done, several days of experimentation. The music room is much bigger and taller and irregularly shaped, the results were even better:
This is why I rate room correction and speakers higher that electronic tweaks.
Yeah, to achieve a curve like that is a great result.
I constantly see people wanting to upgrade a dac or amplifier. Worse is putting up all that ugly room treatment - at least for most home applications with multipurpose / shared rooms.
DSP / room correction is a much cheaper option which will get better results.
Hi all, after much experimenting with different filters, PEQ settings, and dB biases I’ve decided to turn off DSP in Roon. I use Roon primarily with headphones so I don’t use room correction.
I find the DSP module saps air out of the music. The result is a bit muffled and constricted compared to just playing bitperfect to my DAC. I don’t have a very fancy chain…straight out to an Oppo HA-2SE and to a headphone amp.
Am I doing something wrong? I would like to use the DSP–primarily to tweak EQ–but not at the expense of the sound quality degradation I’m getting.
I’m with you all the way. Tried to love upsampling so many times. DSD, upsampled PCM files. Audirvana, now Roon. I can’t tell you how many times I’ve given it a go. I always come back to native - no question about it, in my system. Your description of upsampling is perfect. It sounds ‘filtered’ and flat. Slightly reconstructed, like plastic surgery. I smell engineering/marketing BS on a grand scale. This year’s loudness button, for sure.
That really is the key. You need to try it and see if it works in your system. Although certain dacs can handle certain formats and sample rates, does not mean they sound the best with those. The classic example is the dac marketed as DSD capable but actually performs terribly with DSD.
The only reason to upsample or resample is to achieve the optimum rate for your system. Thinking that blindly upsampling is going to improve the sound is your grand scale BS.
Upsampling is, above everything else, DAC dependent. DSD upsampling can be quite different sounding depending on DAC and isn’t always to everyone’s taste. But if you can hear the difference then that is the least you might expect and proves the theory that upsampling makes a difference. Liking that difference is something else completely!
As far as DSD, the Oppo HA-2SE decodes it well. I consistently find DSD recordings/masters to sound very very good (e.g., Cat Steven’s Tea for the Tillerman) in my setup. The effect I noted below happens also with DSD whether resampling through PCM back to DSD or doing native DSD processing.
I agree, in my opinion, high resolution PCM settings, seem to sound the most accurate, and natural, after comparing all the options. Not a big fan of DSD (Muddy top end, and poorly defined sound-stage), though have heard many SACD’s converted to PCM, sound excellent.
Marking this thread so I can come back and learn about all this.
You can always find the forum’s bookmark feature right under the last post.
You can access your bookmarks by clicking on your profile picture (top right).
No need to post anything just to mark a thread. Also notifications can be turned on, or off should one want to (directly below the Bookmark button or at the right side under the time navigation bar).
