Offline Upsampling

[jtwrace]

Tried your settings and it works!

16/44

 

Converted

[Superdad]

I did the up-sample that I posted with a 16/44 file and it's now shows in JRiver 18432 aiff. What does this mean?

 

18,432 is the kbps bit rate for stereo 24/384KHz.

[Geoffrey Armstrong]

Geoffrey,

From yesturday I intensive think about including DSD into AuI. Of course, DSD encoding must appear in AuI. Now I try find DSD license owner. Also possible variant creating my own algorithm of 1-bit modulation encoding compatible with DSD (DSD is protected trademark).

 

Now I have rights of using DSF files, but algorithm encoding PCM to DSD is protected by this patent https://www.google.ru/patents/WO2006129215A2?cl=en As I know (while not exact) protected patent formula only - algorithm of converting PCM to DSD, but not sigma-delta modulation.

 

Yuri

 

Yuri, I can't comment on the legal or technical issues. It sounds like it could be more interesting to create your own algorithm.

 

There is an attempt here to compare Audiogate, J River and Saracon, doing offline conversion to DSD. Not surprisingly Saracon came out the winner; but the author of this blog questions whether the differences between the 3 would actually be audible.

 

Getting the distortions as low as Saracon would be worth aiming for.

[jaffar]

I compared Redbook to DSD256 in HQP, then Redbook Up-sampled to 352.8 using AuI to DSD256 in HQP and finally Up-sampled to 352.8 file further converted to DSD128 using J River, then played at DSD256 via HQP.

 

The results of all three of these using HQPlayer were so good, I could easily live with any one of them.

 

Thanks for posting the results!

It would be interesting to do this comparison in a blind ABX test (ask a friend to switch the tracks).

I wonder if you can 100% differentiate your favorite upsampling method...

[Jud]

Thanks for posting the results!

It would be interesting to do this comparison in a blind ABX test (ask a friend to switch the tracks).

I wonder if you can 100% differentiate your favorite upsampling method...

 

Please do the test and let us know how you scored.

[Jud]

Yuri, I can't comment on the legal or technical issues. It sounds like it could be more interesting to create your own algorithm.

 

There is an attempt here to compare Audiogate, J River and Saracon, doing offline conversion to DSD. Not surprisingly Saracon came out the winner; but the author of this blog questions whether the differences between the 3 would actually be audible.

 

Getting the distortions as low as Saracon would be worth aiming for.

 

The author of the blog has questioned whether higher sample rates/resolutions make any audible difference; whether different software players do; and whether various other changes do. That's pretty much what he does. Edit: I see what the blog post is about is conversion from PCM to DSD and back, and whether the PCM results are audibly different than the PCM files prior to conversion. That's a little different than quality/audibility of conversion from PCM to DSD.

[Geoffrey Armstrong]

The author of the blog has questioned whether higher sample rates/resolutions make any audible difference; whether different software players do; and whether various other changes do. That's pretty much what he does. Edit: I see what the blog post is about is conversion from PCM to DSD and back, and whether the PCM results are audibly different than the PCM files prior to conversion. That's a little different than quality/audibility of conversion from PCM to DSD.

 

There's another post on his blog where he acknowledges that PCM to DSD conversion yields more pleasing results. He wasn't sure if the results were purely euphonic or otherwise. As always the better the system, the more obvious the improvement in going from PCM to DSD will be …and that goes for the software player too. For me the pecking order starting from the best is as follows: 1 HQPlayer, 2 foo_asio, 3 J River.

 

Please read his subjective listening impressions towards the bottom of that page and note the software and system he used.

 

You can use foo_asio as your output device in J River and take the source all the way to DSD256, if you have a DAC which can support that. It's also more efficient and less demanding on resources than J River's own converter, which surprises me, as it sounds better.

 

This has allowed me to convert internet streams to DSD inline.

 

J River seems to do a great job of offline conversion to DSD128 though. So if this were incorporated into AuI, it would be more a matter of convenience.

[Jud]

Geoff, quick question: Can foo_asio be used as a standalone (separate from JRiver) offline converter?

[Geoffrey Armstrong]

Geoff, quick question: Can foo_asio be used as a standalone (separate from JRiver) offline converter?

 

Unfortunately no. It appears to only be for realtime conversion.

[bogi]

Unfortunately no. It appears to only be for realtime conversion.

 

In foobar it is the same, only real-time.

[audiventory]

Geoffrey, I like your links. Interesting material.

 

"I see what the blog post is about is conversion from PCM to DSD and back, and whether the PCM results are audibly different than the PCM files prior to conversion. That's a little different than quality/audibility of conversion from PCM to DSD."

 

Jud, it (audio comparision) would useful addition to the article.

[bogi]

What if I combine both? Offline upsample to 352.8 and have HQPlayer convert this to DSD256.

 

Now I felt I had the best of both worlds, except I would say the PRAT factor is slightly reduced when more information appears to be added. This reminds me of comparing native DSD to plain unadulterated redbook. It's usually clear the native DSD is richer in information; but often there is still something to be said for plain redbook, in terms of how I respond to the music.

 

My feeling is the up-sampled version, converted to DSD, is closer to native DSD, whereas Redbook converted to DSD walks a line between the two.

 

I came to the same result in Foobar2000. There is one difference between HQPlayer and Foobar: I can do both things in real time directly in Foobar. I use SoX resampler as DSP plugin and foo_sacd. At first, DSP plugins are applied and then foo_sacd is applied. My DAC supports max. DSD128 and I have best results with upsampling by SoX to 352800.

[bogi]

If I wish to do PCM resampling offline (mostly only downsampling for my RockBoxed iPod - imod), I do it in Foobar2000 and I use the Convert option from playlist context menu. It is possible to specify file format and it is possible to apply any installed DSP plugin. So for my iPod I am often applying not only SoX resampler, but also stereo to binaural plugin for headphone listening.

[EuroDriver]

Today I was privileged to hear a very B grade Redbook live recording of Pinkus Zuckerman in 2006 with the NY Philharmonic, upsampled with AuI to 352, then converted by by JRMC to DSD128 and then up sampled on the fly to DSD256 by HQ Payer. The performance was exquiste and glorious. I also listened to the 352 PCM file and that sounded very good already. The conversion to DSD 128, and then upsampling to DSD was the icng on the cake. Simpy stunning that B grade Redbook could sound so fantastic. We then had a reality check in the form of a DSD128 recording of MT Thomas with SF Symphony upsampled to DSD256 wih DSD7 setting, it left Redbook in the dust.

 

Just a few thoughts.

- how much of the improved sound quality is due to better algorithms, and how much is due to the final online digital to analog conversion being carried out with a light PCU processing load, it is very hard to tell ! We have to wait for the Linux drivers for the Exasound, to see what SQ difference splitting the computation from the D>A conversion makes. My guess is that it will make the small ARM playback computer sound ver gu

- when you have sound quality of such a high level, the best sounding balance of resolution, timbre / emotion, PRAT is a work of artistic

 

It's amazing to hear how Geoff has assembled the attributes of so many software players to achieve a sound that leaves me awe struck

[Jud]

EuroDriver - Yes, and hopefully we will get at least one solution that will allow conversion to DSD64, -128, -256 (my DAC's max is -128) completely offline, which might even be better yet. (I have yet to compare upsampling to max res PCM to the same for DSD with my DAC, which should be interesting.)

[EuroDriver]

EuroDriver - Yes, and hopefully we will get at least one solution that will allow conversion to DSD64, -128, -256 (my DAC's max is -128) completely offline, which might even be better yet. (I have yet to compare upsampling to max res PCM to the same for DSD with my DAC, which should be interesting.)

 

I also wonder if HQ Player's NAA option will level the playing field between offline and inline conversion. All of the computational heavy lifting is done in one beast, whilst the delicate act of quietly extracting the analog audio from the bit stream is done in another barebones CPU with a sound friendly OS

[Geoffrey Armstrong]

I came to the same result in Foobar2000. There is one difference between HQPlayer and Foobar: I can do both things in real time directly in Foobar. I use SoX resampler as DSP plugin and foo_sacd. At first, DSP plugins are applied and then foo_sacd is applied. My DAC supports max. DSD128 and I have best results with upsampling by SoX to 352800.

 

Thanks Bogi. Could you provide a link to that SoX re-sampler plug-in for Foobar? I do have a re sampler for Foobar, I'm not sure if it's the same one though.

 

I also like to run streams through foo_asio, such as Qobuz. I can do this using foo_asio as the output device in J River and use J R's live playback feature.

 

Another possibility is to use Bug Head Emperor to up-sample and use J River as the output device for Bug Head, then J River in turn uses foo_asio for its output.

 

This requires some serious CPU horsepower though and this thread is really about offline up-sampling, for people who don't have sufficient CPU for all, or some of, these tricks.

 

…and finally there is another difference between foo_asio and HQPlayer; HQPlayer sounds better!

[Geoffrey Armstrong]

EuroDriver - Yes, and hopefully we will get at least one solution that will allow conversion to DSD64, -128, -256 (my DAC's max is -128) completely offline, which might even be better yet. (I have yet to compare upsampling to max res PCM to the same for DSD with my DAC, which should be interesting.)

 

Jud, I just want to make sure you are aware that you can do this already in J River, 'cos it's not so obvious. Take a file that's in J River, right-click on it and choose Library tools -> convert. The action box appears at the bottom left of J R's window. Here you choose what format you want to convert to, which will be DSD and you can further choose whether DSD64 or DSD128 by clicking on the … for options.

 

You could do this with a file not already in J R's library by right clicking on it in Windows explorer or Mac's finder and choose to open the file with J R. The file will then be accessible by clicking in J R's "Playing Now" Zone in the upper left panel of JR's window.

 

It should work on both Mac and Windows versions of JR and, yes you can perform this on a file already up-sampled by AuI.

 

This is the method I applied in EuroDriver's description above, then I played the resulting file with HQPlayer, using poly-sinc and DSD7.

 

Hope it helps!

[audiventory]

I began builtin to AuI 1-bit audio resolution on 2,8 and 5,6 MHz. Result of the convertion will *.dsd file.

 

As I understand DSD is trademark by Philips. This standard use pulse-dencity modulation. Patented decoder only. I will create own (other than Phillips's algorithm) decoder for converting multibit sequence to on-bit sequence for any sampling rate.

[Jud]

I also wonder if HQ Player's NAA option will level the playing field between offline and inline conversion. All of the computational heavy lifting is done in one beast, whilst the delicate act of quietly extracting the analog audio from the bit stream is done in another barebones CPU with a sound friendly OS

 

Miska has said he thinks NAA makes up at least some of the ground between inline conversion sans NAA and offline. I could actually do that from my upstairs Win desktop through an NAA, but I don't know how practical that arrangement would be - control a remote computer presumably via screen sharing/VNC to have it upconvert to DSD and send the result via Wi-Fi through the NAA - or how it would affect the sonic results. Offline conversion just seems a little more simple to me, and frankly a process that would promote more of the mood I like to be in when enjoying music, rather than having to worry about whether multiple parts of a more complex process will all mesh correctly.

[Jud]

I began builtin to AuI 1-bit audio resolution on 2,8 and 5,6 MHz. Result of the convertion will *.dsd file.

 

As I understand DSD is trademark by Philips. This standard use pulse-dencity modulation. Patented decoder only. I will create own (other than Phillips's algorithm) decoder for converting multibit sequence to on-bit sequence for any sampling rate.

 

Thanks, Yuri, this is very encouraging and welcome news. When you say the result will be a *.dsd file, do you mean rather than .dsf or .dff (which, as far as I know, are the file types my software players will play - .dsf preferred because it is better for metadata)?

[Jud]

Jud, I just want to make sure you are aware that you can do this already in J River, 'cos it's not so obvious. Take a file that's in J River, right-click on it and choose Library tools -> convert. The action box appears at the bottom left of J R's window. Here you choose what format you want to convert to, which will be DSD and you can further choose whether DSD64 or DSD128 by clicking on the … for options.

 

You could do this with a file not already in J R's library by right clicking on it in Windows explorer or Mac's finder and choose to open the file with J R. The file will then be accessible by clicking in J R's "Playing Now" Zone in the upper left panel of JR's window.

 

It should work on both Mac and Windows versions of JR and, yes you can perform this on a file already up-sampled by AuI.

 

This is the method I applied in EuroDriver's description above, then I played the resulting file with HQPlayer, using poly-sinc and DSD7.

 

Hope it helps!

 

Thanks Geoff, I appreciate the info. At some point I may try this (other than the part where you play the resulting file with HQPlayer ).

[bogi]

Thanks Bogi. Could you provide a link to that SoX re-sampler plug-in for Foobar? I do have a re sampler for Foobar, I'm not sure if it's the same one though.

 

Geoffrey, I downloaded it from here: A new resampler DSP for foobar2000 - Hydrogenaudio Forums

[bogi]

EuroDriver - Yes, and hopefully we will get at least one solution that will allow conversion to DSD64, -128, -256 (my DAC's max is -128) completely offline, which might even be better yet. (I have yet to compare upsampling to max res PCM to the same for DSD with my DAC, which should be interesting.)

 

I found this SW has to be able to do that, but I don't know how much it costs:

Weiss Saracon SARACON | Weiss Engineering Ltd.

[jtwrace]

I found this SW has to be able to do that, but I don't know how much it costs:

Weiss Saracon SARACON | Weiss Engineering Ltd.

 

$2250 here in the USA