Offline Upsampling

[SwissBear]

Hi SwissBear,

Bing number items in input file list theoretically can cause problem (memory issue?). I will try model your situation (300 albums x 15 tracks, each in separate directory).

If possibly wait 2-3 days, I will model it after end of the new site design. If it is urgent trouble (write me) I will begin it today-tomorrow (need prepare library model AIFF 16/44).

Hi Yuri,

Absolutely no hurry as far as I am concerned. As I wrote, the product is now working according to my expectations. Bugs can be looked after when appropriate.

[audiventory]

Hi SwissBear,

 

Thank you for waiting. I will write when to begin the test.

[sdolezalek]

Yuri: I have sent you an email at [email protected] inviting you to share a DropBox folder that contains a WORD version of the User Manual that I have started to edit. So far, I have only made it through the first 5 or 6 pages. Let me know if you want to invite others who can also edit this file or whether you want to move it to a different sharing application.

[audiventory]

Sdolezalek,

 

Thank you. I uses dropbox. And received your shared directory.

 

I heard what googledocs allow works under shared documents too. But havn't opportunity use it while.

 

You can share the directory how you consider. As I wrote above the manual under free license now that will allow free contribute and share the manual. Authors must listed there. I create license text later (simple, like flac's license).

[sdolezalek]

Sdolezalek,

 

Thank you. I uses dropbox. And received your shared directory.

 

I heard what googledocs allow works under shared documents too. But havn't opportunity use it while.

 

You can share the directory how you consider. As I wrote above the manual under free license now that will allow free contribute and share the manual. Authors must listed there. I create license text later (simple, like flac's license).

 

Thanks Yuri. If anyone else wants to participate in helping to draft a clearer English version of the User Manual send me a PM with your email and I'll permission you with editor access in Dropbox.

[pipis2010]

This is all very interesting I have to say. I found this thread by accident, but I am definitely going to download the software to try!!!

I use Daphile OS with an iFi Nano DSD and am curious of the DSD256 upsampling...

 

 

Sent from my iPhone using Computer Audiophile mobile app

[SwissBear]

Thanks Yuri. If anyone else wants to participate in helping to draft a clearer English version of the User Manual send me a PM with your email and I'll permission you with editor access in Dropbox.

Hi sdolezalec,

I'm willing to help, even though I'm not an English native speaker/writer.

[audiventory]

Sdolezalek, SwissBear,

 

I'm very appreciated your assistance.

[tranz]

Does the demo version allow for all formats and filters even though it warns that the higher resolutions are only available for Pro? Thanks

 

A side note is that browser protection/ anti-virus throw up a bunch of warnings downloading and installing it.

[Jud]

Does the demo version allow for all formats and filters even though it warns that the higher resolutions are only available for Pro? Thanks

 

A side note is that browser protection/ anti-virus throw up a bunch of warnings downloading and installing it.

 

Don't remember regarding resolutions, someone else has to answer or you can see for yourself.

 

Re warnings, I've had no problems. By the way, may I interest you in breast enlargement?

[audiventory]

Hi Tranz,

 

Does the demo version allow for all formats and filters even though it warns that the higher resolutions are only available for Pro? Thanks

 

All filters/resolutions available. Except PCM 32 bit. Possibly, I will open it for DEMO in future.

DSD filters mute 2 second silence in middle of output track.

All warnings for better understanding of editions abilities.

 

 

A side note is that browser protection/ anti-virus throw up a bunch of warnings downloading and installing it.

 

Firewall should warn about new attempts of connections (for downloading of codecs) in learning mode.

 

It is fully correct. Don’t worry about it.

[tranz]

Sweet! Thank you.

[audiventory]

Hi Yuri,

Absolutely no hurry as far as I am concerned. As I wrote, the product is now working according to my expectations. Bugs can be looked after when appropriate.

 

Hi SwissBear,

 

I have created model directory structure AIFF 44/16 x 20 tracks 300+ albums.

 

It caused interface lags. I suppose, it depend on available memory size first due used bigger number of list elements. Also inter-thread (parallel processings) management issue being possibly.

 

Looks like processing core limitation in the AuI's settings don't influence (significantly?) to the lags of graphical user interface (GUI).

 

As instant decission possible reduce number of files in AuI's input file list.

 

I will thinlk how unload GUI for fixing its lags. Here need significant (as looks now) re-working (enought time need). First me need thinks how do it simpler in current AuI's structure. While I will learn it and will notify you about results.

[SwissBear]

Hi SwissBear,

 

I have created model directory structure AIFF 44/16 x 20 tracks 300+ albums.

 

It caused interface lags. I suppose, it depend on available memory size first due used bigger number of list elements. Also inter-thread (parallel processings) management issue being possibly.

 

Looks like processing core limitation in the AuI's settings don't influence (significantly?) to the lags of graphical user interface (GUI).

 

As instant decission possible reduce number of files in AuI's input file list.

 

I will thinlk how unload GUI for fixing its lags. Here need significant (as looks now) re-working (enought time need). First me need thinks how do it simpler in current AuI's structure. While I will learn it and will notify you about results.

Hi Yuri,

 

Thanks for the update. I have adapted my workflow to the existing application. No worry.

[SwissBear]

Hi Yuri,

 

A short question about sampling rates. My DAC is accepting a max of 24bit/192kHz.

Do you recommend that I systematically upsample to 24/192 with your software or should I make tests to check if 24/96 is better ?

 

Thanks,

[audiventory]

Hi SwissBear,

 

Maximal sample rate value don't guarantee the best sound.

 

Better way is checking of different sample rates.

 

Theoretically, there are may be difference for certain instance of model of DAC even.

 

Best regards,

Yuri

[tranz]

First off, thank you for the trial version. The tool was easy to use.

 

This is just one person's feedback on one system. I tried a CD rip up to 24/384 and DSD256 with default settings.

 

The DSD256 file did not play back.

 

The DSD128 sounded fine except not as good as the original. Especially on guitar and vocals the space and sparkle was lowered/deadened.

 

The 24/384 file added a level of noise, although slight, that made it not as good as the original.

 

This was on a discrete ladder DAC, so I guess the experience will be DAC dependent.

 

Cheers

[Jud]

First off, thank you for the trial version. The tool was easy to use.

 

This is just one person's feedback on one system. I tried a CD rip up to 24/384 and DSD256 with default settings.

 

The DSD256 file did not play back.

 

The DSD128 sounded fine except not as good as the original. Especially on guitar and vocals the space and sparkle was lowered/deadened.

 

The 24/384 file added a level of noise, although slight, that made it not as good as the original.

 

This was on a discrete ladder DAC, so I guess the experience will be DAC dependent.

 

Cheers

 

1 - What DAC?

 

2 - Did you measure playback loudness to determine it was equal in each case? (AudioTools dB graphing app for iPhone can do this.)

[audiventory]

Hi Tranz,

 

Thank you for so detailed feedback. It is very important and interesting information.

 

1. What operation system you use (Win or Mac)?

 

 

I tried a CD rip up to 24/384 and DSD256 with default settings.

 

The DSD256 file did not play back.

 

I think, need check supporting of DSD256 mode.

 

2. What software and hardware you used for the testing playback?

 

The DSD128 sounded fine except not as good as the original. Especially on guitar and vocals the space and sparkle was lowered/deadened.

 

Bit perfect hypothesis

 

If playback is not bit-perfect, can be difference due inline conversion in playback software and hardware.

 

If playback is bit-perfect, possibly DSD128 circuits work other way than 16/44.

 

I always recommend to seek «best sounding» resolution (sample rate / bit depth / PCM | DSD) in available range for each system (playback software+hardware).

 

It is one of reasons why technically impossibly comparing DSD vs. PCM or 44 vs. 192.

 

 

3. Did you compare CD audio hardware player (original) with playback via computer?

4. What DAC used for listening of CD?

 

 

The 24/384 file added a level of noise, although slight, that made it not as good as the original.

 

DAC internal oversampling hypothesis

 

If DAC provide internal oversampling can appear new artefacts.

 

I think, in 384 mode DAC’s output analog filter have more wide band. What influence to general artifact level.

 

AuI can provide non-distorted signal in pass-band (0 … 20 kHz for optimized mode, 0 … 1/2 sample rate in non-optimized mode),

and suppressing noise/artefacts in stop band for providing level about -146 dB for 24 bit resolution PCM.

For source is 16/44 output noise level may be about -100 ... -120 dB.

It under our control.

 

You noted discrete ladder DAC. Possibly used system will work better in 16/384 or 16/352 mode.

 

 

 

2 - Did you measure playback loudness to determine it was equal in each case? (AudioTools dB graphing app for iPhone can do this.)

 

Level difference hypothesis

 

 

As said Jud, possibly different perceived quality for different levels in different modes. Even «invisible» 1 … 2 dB difference can be significant for comparing of the quality.

 

AuI’s DSD output have -0.4 dB gain. For providing stability DSD modulator to signals with highest energy.

 

Usually the gain applied in DSD modulators as -6 dB.

 

Possibly compensate -0.4 dB via AuI's manual gain control in commercial editions.

 

 

 

DAC can use different gain (sometimes switched) for DSD and PCM mode.

 

 

Distortions due overloading hypothesis

 

Also due oversampling of musical signal possibly overloading. I know real cases when CD upsampling caused overloads and distortions.

 

Level normalization automatically decreased level for avoiding of overloads. Read more here How Maximize Loudness Audio File and Avoid Overload

 

In DEMO AuI level normalization available for ISO extracting mode only.

 

In commercial editions possibly manual decreasing gain for avoiding of the overload. I think, enough -1 … -3 dB gain.

 

 

 

Currently computer audio systems enough sophisticate with many hidden at first look details.

 

Best regards,

Yuri

[tranz]

...1. What operation system you use (Win or Mac)?

 

I think, need check supporting of DSD256 mode.

 

2. What software and hardware you used for the testing playback?

...

If playback is not bit-perfect, can be difference due inline conversion in playback software and hardware.

...

It is one of reasons why technically impossibly comparing DSD vs. PCM or 44 vs. 192.

 

3. Did you compare CD audio hardware player (original) with playback via computer?

 

4. What DAC used for listening of CD

 

...

 

Hi Yuri,

 

Thanks for the detailed reply.

 

The DAC is an MSB Analog using a Renderer (LAN input) module. (Direct to I2S using the DAC's femto clock) No CD player was used.

 

Files are ripped using EAC or DBPoweramp in secure and AccurateRip verified mode. Bit perfect playback of the DAC and source was tested beforehand.

 

A Windows 8.1 pro laptop was used to run your program.

 

Playback is done through MinimServer / BubbleUPnP on a NAS. Optical LAN is in place.

 

I used a dB meter to check volumes, and there was a slight volume difference between DSD and PCM files, but that was not the cause.

 

Here is some info on the internal DAC filtering: 32xFIR with 6000 taps. They state the noise floor as <145dB.

 

http://www.msbtech.com/products/32xdf.php?Page=dac5signature

 

I will try 16bit, but in the past I have compared 24 and 16bit versions of rips with no perceived impact due to the zeroes padding.

 

DSD256 playback was tested using purchased files to make sure it was not an issue with the DAC. I have noticed in the past that material recorded in DSD originally always sounds better than PCM to DSD conversions on this DAC.

 

Cheers

[Jud]

Hi Yuri,

 

Thanks for the detailed reply.

 

The DAC is an MSB Analog using a Renderer (LAN input) module. (Direct to I2S using the DAC's femto clock) No CD player was used.

 

Files are ripped using EAC or DBPoweramp in secure and AccurateRip verified mode. Bit perfect playback of the DAC and source was tested beforehand.

 

A Windows 8.1 pro laptop was used to run your program.

 

Playback is done through MinimServer / BubbleUPnP on a NAS. Optical LAN is in place.

 

I used a dB meter to check volumes, and there was a slight volume difference between DSD and PCM files, but that was not the cause.

 

Here is some info on the internal DAC filtering: 32xFIR with 6000 taps. They state the noise floor as <145dB.

 

MSB Digital Filters

 

I will try 16bit, but in the past I have compared 24 and 16bit versions of rips with no perceived impact due to the zeroes padding.

 

DSD256 playback was tested using purchased files to make sure it was not an issue with the DAC. I have noticed in the past that material recorded in DSD originally always sounds better than PCM to DSD conversions on this DAC.

 

Cheers

 

Hi tranz,

 

I'm very curious about this stuff and figure you are too, so if you are agreeable, I would like to try to do the volume equalization available in the paid version of AuI and also try out Yuri's minimum phase filter to see if they make any difference for you. Please PM me and we can determine what track(s) you want to use for testing.

 

To do the volume equalization, I would like to know with a fair amount of precision (to ~1/10dB if possible) what the volume difference is between the files you converted with the demo version of AuI and those converted in-DAC. If the dB meter you used moves too quickly to tell this, you might try out AudioTools or something similar with a dB graphing function that allows you to preserve graphs of loudness levels to compare at leisure.

[audiventory]

Hi Tranz,

 

Thank you additional information.

 

Playback is done through MinimServer / BubbleUPnP on a NAS. Optical LAN is in place.

…

DSD256 playback was tested using purchased files to make sure it was not an issue with the DAC.

 

I just convert FLAC 44 to DSD 256.

 

Unfortunately I can’t fast start the minim server (some error displayed). May be I can launch it later. But I found nowhere about minim’s DSD256 ability.

 

Foobar playback successfully.

 

Tascam’s Hi-Res editor can open and playback converted file.

 

If in Settings window > Metadata tab > Compability metatags with iTunes is checked, may be troubles with some software (iTunes use non-standard format of metadata). Need uncheck this switch.

 

 

1. Purchased files was DSD256?

 

2. It was DSF or DFF?

 

3. It was played back via current minim server?

 

 

I have noticed in the past that material recorded in DSD originally always sounds better than PCM to DSD conversions on this DAC.

 

Here compared two different full systems (from ADC to DAC).

 

Final pro-level conversion PCM (by AuI or other manufacturers) to DSD can be insignificant in big number of previous processing stages.

 

There no additional information «between bits». It simply check via [test file PCM] > DSD > [result PCM] conversion and checked with RMAA software as example. Or via manual analyzing with other software.

 

I will try 16bit, but in the past I have compared 24 and 16bit versions of rips with no perceived impact due to the zeroes padding.

 

Except developers nobody know how DAC process inside. Adding by zeroes is not the best way. AuI convert input to 64-bit float. Process it in this format. And round to 16 bit for 16 bit output formats.

Without processing AuI produce output fully identical to input.

 

If 16/44 converted to 16/384 (processed), as example, have sense try dithering ON. If will not difference or will worse, need dither OFF. Result depend on source stuff.

Also can be adjusted intensity in Settings window > General tab > Dithering mask … slider.

[Jud]

As a ladder DAC, does the MSB analog DAC sigma-delta modulate PCM input? Another way of asking pretty much the same thing is, how does this DAC internally handle PCM vs. DSD input?

[audiventory]

Hi Yuri,

 

Thanks for the update. I have adapted my workflow to the existing application. No worry.

 

Hi SwissBear,

 

After experiments under modeled 300+ directories I found 3 place of dalays:

 

1. When files opening during seek source files in directory, absent codecs, etc.

This delay is interface hanging after clicking Open button in Chose directory dialog. In my system (Mac + NAS) it about 1-2 minutes.

 

2. After pushing Start button begin check for presence CD, ISO, CUE etc. - all what demands additional processing before converting. Pre-processing performed in separate thread and don't hang graphical interface (GUI) of AuI ConverteR. But detection of things that need pre-process performed directly after pushing the Start button. And hang GUI.

 

3. Log processing strange work (I suppose, used program library issue). For big number of used files it can input in high resource consuming mode.

And even AuI stay in pause mode, log processing consuming same resources anyway. It cause slugging during conversion that you view.

 

I found way how fix it and released it now. Currently in pause mode AuI consume resources in idle mode level (it's correct). And don't cause the slugging (at my system of course).

 

After several tests I will send you beta version for checking on your system. I hope slugging during conversion will fixed on your system too

 

First two issues demands more deep re-working and I will think how release it simpler (for development and using).

[SwissBear]

Hi Yuri,

Thanks very much for the update and the corrections.

Hope to be able to test later this week.