Hi Matt,
Not sure I understand why.
So Audirvana does a conversion from PCM to DSD, streams it via DLNA to a DAC like Katana, that will convert it back to PCM before the analog output stage?
If I understood you correctly then I believe there are too many conversions, risk deterioranting the data.
My understanding of DSD (that uses Pulse Density Modulation) is that the audio quality will benefit from a much simple analog output stage but maybe the conversion from PCM to DSD in the DAC chip is more complex.
The PSAudio DS does the conversion to DSD in a FPGA.
The RPi can not do this because conversion to DSD needs a lot of computing power.
As I mentioned the AK4137 is not a very good sounding solution.
Most DAC chips do a PCM to DSD conversion anyway, but in most cases software conversion to DSD is superior so the DAC chip has only to do the conversion from DSD to analog.
Matt
If I may, a topic for discussion and perhaps for a future DAC
I recently read that DSD signal is way more easy to convert into analogue audio signal then a PCM signal. The output stage is less complex, resulting in better sound quality.
Don't know if you agree.
Given the above, there is at least a high end audio manufacturer that I know, converts the PCM into DSD (and not the other way around) before the analogue output stage.
Any thoughts about it?
Apologies that this is off topic but the question was asked. The point about decoding DSD is that you don't actually need a DAC (in the conventional sense of a DAC chip based solution), just a LP filter. You might want to check out these threads;
For a basic solution;
The Best DAC is no DAC
and a more refined, better sounding and better behaving solution;
Signalyst DSC1
I've done quite a bit of experimenting with DSD decoding and the sound quality is sublime and the recent DSC2 implementations by DIY Audio member 'ppy' are now my absolute benchmark and I don't anticipate adopting a conventional DAC solution again.
The downside of DSD is the paucity of recordings so the only practical solution is to convert/upsample PCM to DSD but to do that optimally you need something like HQ Player and a powerful PC - and yes, DSD that has been competently converted/upsampled from PCM sounds better than the PCM played back via a conventional DAC. All just my opinion of course, YMMV.
Then I guess, like Matt said, the RPi doesn't have enough computing power for such conversion.
Then I guess, like Matt said, the RPi doesn't have enough computing power for such conversion.
Nowhere near enough processing power for the conversion but enough to serve as an endpoint, such as an HQ Player NAA (network audio appliance), that renders the DSD data stream. One of the chaps on the Signalyst DSC1 thread I linked to is using a Beaglebone Black in such a role.
Also, it is my understanding that many of the conventional DAC chips that claim to be DSD capable actually convert DSD to PCM internally before decoding it.
When enabling hardware volume control some chips are converting DSD back to PCM for performing volume control.
Some chips like AKM have a direct DSD path without volume control.
Matt
Yes, that's why I said "many of the conventional DACs". I have tried some DSD capable AKM DACs and they don't sound as good as the Signalyst DSC2, nor does my ESS9018 Buffalo 3SE.
Pobably. Hardware wise the issue might be board space but perhaps two mono boards stacked (and without the output transformers) might be doable. Would also need the right software on the RPi as you must generate an effective mute flag as well as the DSD data and clock outputs. Probably simpler to just interface the RPi to one of the DSC2 boards I linked to - as I mentioned somebody is using a BBB in this way (with Botic software and via a Cronus reclocker).
We have been experimenting with the DSC1 for a while now. However a proper design takes anywhere 5-10 months.
Yes I did. Sounds good (to say the least ). What I have in the office though, measures worst than Katana but sounds "almost" as good.
Alternative architectures (like r2r , dsc1) definitely bring a different sound . A great sound.. Also transformers have s sig sound , as well as tube stages and every opamp there is .Everyone has a taste for one over another.
Katana is based on a Delta Sigma IC. Millions of those are sold and lots of money goes into research and development . With the proper implementation on clocks (they are very sensitive to jitter) and output stage , the question becomes less what IC was used , or what architecture,.
The only question is: How does it sound.
As I said since the beginning...Katana was designed for sound quality only. THD+ N numbers are meaningless to us, just a way to prove we know what we are doing (yes they do help somewhat SQ)
Lots of DIY people have an opinion(and preference) on the opamp sound , but they never heard discrete op amps . Sparkos Labs with Allo, we hope to change that perception . We hope you will understand what was lost (SQ) by integrating circuits..
They (sparkos) dont sound like opamps , they don't sound like transformers or tubes . They don't color the sound but they do something ..they free it.
I will stop because words cannot describe, adequately, sound.
Soon Katana will be available (in small qty for first run). Sit back , stream your favorites and get lost in another world.
Thats our goal with Katana.
Within reason, I can't get very excited about the numbers and, as the vendor of Katana, you'll hopefully forgive me if I discount your perspective on the sound quality comparison.
The question I was asking was whether there is actually a potential DSD product in preparation but I guess you need to get Katana launched onto the market before giving too much away about what might come next?