dff to dsf conversion in Jriver

[Skeptic]

I dont think HQ Player has any PCM conversion when upsampling DSD. I recall Misjka saying so..

 

What do you mean by PCM, multi-bit SDM or DSD-Wide?

I suppose it would be possible to perform the filtering in multi-bit SDM/DSD-Wide/2.8MHz PCM and then convert to 2x/4x DSD.

 

My point was more that you have to convert it away from being native 1xDSD to filter out the high frequency noise before you convert it to 2x/4x DSD rates - that you can't just go straight from 1xDSD to 2x DSD without that filtering step.

At least not without introducing a lot of high frequency noise on the output.

[mlgrado]

Indeed, DSD rates higher than 128 do divert to PCM, since the real way to go higher than 128 with DSD is not quite there (yet).

 

 

That is the first I have heard this. Pardon me for being contrary, but I believe this is somewhat incorrect. Jriver will bitstream a native DSD file as DSD, regardless of it being 64fs, 128fs, or 256fs if you use ASIO Native.

 

I am playing back a Native DSD256 file at the moment, downloaded from NativeDSD.com. The 'bitperfect' light is on as expected. The audio path is reported as being "Direct", and "No Changes" are being made from input to output.

 

I can do this with native ASIO streaming, or with DoP, since my Dac supports up to 768khz PCM input.

 

Now, what I think you may be talking about, is the inability to stream via DoP at DSD256. And indeed, I cannot do this using the Jriver DoP engine. But if I use a third party engine, such as ASIO proxy, I can indeed use DSD256 via DoP. My DAC confirms this, as it indicates it is receiving a sample rate of 705khz, exactly as it should for DSD256 via DoP.

 

 

 

[ted_b]

Migrado, You are missing the OP's point. The discussion is not about what JRiver does when bitstreaming DSD or its higher fs rates, but it's about what JRiver does when one tries to simply innocently change DSD formats from say DSF to DFF, or upsample any DSD format to DSD128, for example. DSD conversion in JRiver ALWAYS uses PCM, so for those that want to go redbook to DSD128, fine (cuz PCM is involved regardless). But for those that want to go DSD64 to DSD128 that step ALSO goes through PCM. As does DSF to DIFF. The JRiver team is being transparent about this and does not recommend using their tools to upsample DSD or change DSD formats. Instead use Audiogate, for example.

[mlgrado]

I dont think HQ Player has any PCM conversion when upsampling DSD. I recall Misjka saying so..

 

What do you mean by PCM, multi-bit SDM or DSD-Wide?

 

You are correct. There is no conversion to 'traditional' PCM in the up or down sampling. The sample rate is always at DSD rates. Which means the filtering never approaches anything like typical decimation.

 

You know when places like DSDFile offer both DSD64 and DSD128 versions of a file? Same deal. The DSD64 is likely sourced from the DSD128. Downsampled. In the process of filtering, yes, there is multiplication to a multi-bit form. But the sample rate never drops below 2.8mhz before being remodulated to 1 bit.

 

This is consistent across the DSD 'universe'. It is called DSD-wide some places.

[mlgrado]

Okay, got it. Sorry.

 

 

Migrado, You are missing the OP's point. The discussion is not about what JRiver does when bitstreaming DSD or its higher fs rates, but it's about what JRiver does when one tries to simply innocently change DSD formats from say DSF to DFF, or upsample any DSD format to DSD128, for example. DSD conversion in JRiver ALWAYS uses PCM, so for those that want to go redbook to DSD128, fine (cuz PCM is involved regardless). But for those that want to go DSD64 to DSD128 that step ALSO goes through PCM. As does DSF to DIFF. The JRiver team is being transparent about this and does not recommend using their tools to upsample DSD or change DSD formats. Instead use Audiogate, for example.

[ted_b]

You are correct. There is no conversion to 'traditional' PCM in the up or down sampling. The sample rate is always at DSD rates. Which means the filtering never approaches anything like typical decimation.

 

You know when places like DSDFile offer both DSD64 and DSD128 versions of a file? Same deal. The DSD64 is likely sourced from the DSD128. Downsampled. In the process of filtering, yes, there is multiplication to a multi-bit form. But the sample rate never drops below 2.8mhz before being remodulated to 1 bit.

 

This is consistent across the DSD 'universe'. It is called DSD-wide some places.

 

Yes,. in the case of DSDFile (Opus 3 analog to DSD128 transfers) Jan uses the DSD128 files as source then downsamples to DSD64 for those versions. Net/net, if you can play them buy his DSD128 files; they are the family jewels.

[mlgrado]

Yes,. in the case of DSDFile (Opus 3 analog to DSD128 transfers) Jan uses the DSD128 files as source then downsamples to DSD64 for those versions. Net/net, if you can play them buy his DSD128 files; they are the family jewels.

 

Yes, I agree. The 128 files and the extra two dollars or so, is the way to go.

 

Any DSD, in my opinion, is best at its native recorded rate.

[Miska]

Yes,. in the case of DSDFile (Opus 3 analog to DSD128 transfers) Jan uses the DSD128 files as source then downsamples to DSD64 for those versions. Net/net, if you can play them buy his DSD128 files; they are the family jewels.

 

IMO, it is always worth buying the "original" non-conversion whenever possible. One can perform conversion at playback time if current equipment cannot play it at the moment. I just see buying content in original format as future-proof investment.

 

Anyway I have number of different DACs, not all can play DSD128, but many can. For those that cannot I can still convert on-the-fly to whatever is best supported format.