Jump to content
IGNORED

Offline Upsampling


Jud

Recommended Posts

DFF to FLAC Converter for Mac OSX, Windows.

Features, Configurations, Quick start manual

http://samplerateconverter.com/dff/dff-flac-converter

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
  • 3 weeks later...

Standard audio sample rates are based on 44.1 and 48 kHz. Somebody think that multiple resampling (integer multiplication/division) have lesser distortions than non-multiple one. Read below about difference between both resampling types http://samplerateconverter.com/educational/resampling

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
  • 1 month later...

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment

Who else has tried the Xivero? Comparisons ...?

macmini M1>ethernet / elgar iso tran(2.5kVa, .0005pfd)>consonance pw-3 boards>ghent ethernet(et linkway cat8 jssg360)>etherRegen(js-2)>ghent ethernet(et linkway cat8 jssg360) >ultraRendu (clones lpsu>lps1.2)>curious regen link>rme adi-2 dac(js-2)>cawsey cables>naquadria sp2 passive pre> 1.naquadria lucien mkII.5 power>elac fs249be + elac 4pi plus.2> 2.perreaux9000b(mods)>2x naquadria 12” passive subs.

Link to comment

It does dsd64 upsampling to dsd256 & dsd512?

macmini M1>ethernet / elgar iso tran(2.5kVa, .0005pfd)>consonance pw-3 boards>ghent ethernet(et linkway cat8 jssg360)>etherRegen(js-2)>ghent ethernet(et linkway cat8 jssg360) >ultraRendu (clones lpsu>lps1.2)>curious regen link>rme adi-2 dac(js-2)>cawsey cables>naquadria sp2 passive pre> 1.naquadria lucien mkII.5 power>elac fs249be + elac 4pi plus.2> 2.perreaux9000b(mods)>2x naquadria 12” passive subs.

Link to comment
8 hours ago, jamesg11 said:

It does dsd64 upsampling to dsd256 & dsd512?

 

1 hour ago, bodiebill said:

Yes, amongst other formats. According to their website: 
Output Audio Formats: WAV, AIFF, FLAC and DSD 
Input Audio Formats: WAV, AIFF, FLAC, ALAC, DSD and MP3
Supported DSD Sample Rates: DSD64, DSD128, DSD256 and DSD512

But the way it converts DSD is, well, questionable.

 

When you convert DSD64 to a higher DSD rate the resulting file contains the noise of the initial DSD64 source and the additional noise of the target DSD rate (see attachment #01 - DSD64 converted to DSD128 with XiSRC).

 

Likewise, when you convert DSD64 to, say, 176.4kHz the processed file still contains the noise of the DSD64 source (see attachment #02 - DSD64 converted to 176.4kHz with XiSRC).

 

So obviously there's no lowpass filtering implemented for DSD conversions ...

 

01 XiSRC DSD64-DSD128.jpg

02 XiSRC DSD64-PCM.jpg

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment

double post ...

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
4 minutes ago, bodiebill said:

Too technical for me, I only used my ears. But maybe ask Xivero?

No need to ask.

If the above is too technical, I'll make it simple: don't use this software to convert DSD source files (to any target file).

 

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
32 minutes ago, bodiebill said:

First you asked, then you used the word 'questionable', and now you know for sure? How do you know for sure?

And how do other converters (AuI, JRiver, Tascam) do this?

I did not ask... I simply provided info about how XiSRC handles DSD conversions.

 

With "questionable" I was referring to the way XiSRC converts DSD files.

"Questionable" in the sense of "dubious"... I'm no native speaker, so maybe "questionable" was misleading (?).

 

The graphs above clearly show how XiSRC converts the files.

 

AuI first applies an appropriate lowpass filter to the source file and then converts the actual music content to the target file format (in case the target is a higher DSD rate AuI's modulator will add appropriate shaped noise to the target file).

See attachment and compare it to the first attachment in my post above (same source file / DSD64 to DSD128 conversion).

Don't know JRiver and TASCAM.

 

aui dsd.jpg

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
8 minutes ago, bodiebill said:

Would SiSRC's conversion from PCM to DSD on the other hand be OK?

Yes, absolutely!

 

attached 44.1kHz to DSD128.

The lowpass filter removes aliasing at the nyquist limit of the source file (so at 22050Hz in this case) effectively and the DSD modulator applies appropriate noise shaping.

XiSRC PCM DSD128.jpg

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment

Exc info, c of a.

 

I'm trying out Aul now, via the trial. A lot of $ to weigh up, but no point in accepting the sub-standard.

My aim is to mainly upsample my sacd isos. So, I'm up for the full Aul costing ...

macmini M1>ethernet / elgar iso tran(2.5kVa, .0005pfd)>consonance pw-3 boards>ghent ethernet(et linkway cat8 jssg360)>etherRegen(js-2)>ghent ethernet(et linkway cat8 jssg360) >ultraRendu (clones lpsu>lps1.2)>curious regen link>rme adi-2 dac(js-2)>cawsey cables>naquadria sp2 passive pre> 1.naquadria lucien mkII.5 power>elac fs249be + elac 4pi plus.2> 2.perreaux9000b(mods)>2x naquadria 12” passive subs.

Link to comment
7 hours ago, jamesg11 said:

Exc info, c of a.

 

I'm trying out Aul now, via the trial. A lot of $ to weigh up, but no point in accepting the sub-standard.

My aim is to mainly upsample my sacd isos. So, I'm up for the full Aul costing ...

I would also compare SoX.

XLD contains the SoX resampler/modulator which is a well regarded tool by many (the actual DSD modulator is also contained in Audirvana Plus for example).

AFAIK it can also handle sacd isos (not 100% sure, though).

XLD conversions from PCM to DSD are also much faster than AuI (AuI is super slow on everything...).

And it's free: https://sourceforge.net/projects/xld/

 

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
13 hours ago, copy_of_a said:

I would also compare SoX.

XLD contains the SoX resampler/modulator which is a well regarded tool by many (the actual DSD modulator is also contained in Audirvana Plus for example).

AFAIK it can also handle sacd isos (not 100% sure, though).

XLD conversions from PCM to DSD are also much faster than AuI (AuI is super slow on everything...).

And it's free: https://sourceforge.net/projects/xld/

 

Thanks for this suggestion; will follow up today - it's been my go-to for cd ripping.

We need some of your analysis graphs!

macmini M1>ethernet / elgar iso tran(2.5kVa, .0005pfd)>consonance pw-3 boards>ghent ethernet(et linkway cat8 jssg360)>etherRegen(js-2)>ghent ethernet(et linkway cat8 jssg360) >ultraRendu (clones lpsu>lps1.2)>curious regen link>rme adi-2 dac(js-2)>cawsey cables>naquadria sp2 passive pre> 1.naquadria lucien mkII.5 power>elac fs249be + elac 4pi plus.2> 2.perreaux9000b(mods)>2x naquadria 12” passive subs.

Link to comment

XLD - 44-1kHz to DSD128 (Noise Shaping Order: 6):

 

XLD PCM-DSD128.jpg

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
On 24.8.2017 at 11:19 AM, copy_of_a said:

I'll make it simple: don't use this software to convert DSD source files (to any target file).

 

On 24.8.2017 at 10:35 AM, copy_of_a said:

So obviously there's no lowpass filtering implemented for DSD conversions ...

regarding XiSRC.

 

I'm testing several sample rate converters in the DSD -> PCM szenario. I'm not an expert in this topic, so let me visualize my results for your comments. The screenshots are produced by XiVero's MusicScope.

 

1. Original DFF with musical content up to about 30 kHz and subsequent characteristic DSD noise:

original.dff_report.thumb.jpg.a973cec89feedf1cbc8588ab265bb07c.jpg

 

2. DFF converted to PCM (24/88.2) via foobar2000 and foo_input_sacd plugin:

foobar-foo-input-sacd-88.flac_report.thumb.jpg.579445f016f00abf781a1a397bea9fd8.jpg

 

3. DFF converted to PCM (24/96.0) via foobar2000 and SoX-based foo_dsp_resampler:

foobar-sox-foo-dsp-resampler-96.flac_report.thumb.jpg.772fc6778b77fd88a137179ccf6fcbba.jpg

 

4. DFF converted to PCM (24/96.0) via mansr's DSD enhanced SoX:

sox-dsd-96.flac_report.thumb.jpg.75edb4dff51837d85e0bb8efe5c1c02c.jpg

 

5. DFF converted to PCM (24/96.0) via TEAC's Hi-Res Editor (same as TASCAM's):

TEAC-96.wav_report.thumb.jpg.2cf4b177c4423a5815146273a35a19b7.jpg

 

6. DFF converted to PCM (24/96.0) via XiSRC:

XiSRC-96.flac_report.thumb.jpg.e6522e6c74cc0931002ab61dbad6f173.jpg

 

None of the five resamplers show a lowpass filtering which is IMHO conformant to Sony's recommendation to apply a filter with a cut-off frequency of about 50 kHz.

 

My questions:

1. Do the screenshots provide any insights about the quality of the resamplers?

2. @copy_of_a: Do you keep your verdict about XiSRC and the conversion of DSD material? If yes, why?

 

Link to comment
9 minutes ago, klassikmann said:

1. Do the screenshots provide any insights about the quality of the resamplers?

 

Quality of resamplers is artefacts (non-linear distortions, aliases) and noise level.

 

Also DSD modulators may be compared by overload steadiness. Because overload cause unstable state of the modulator (silence, tone, etc.), that can be restored after reset the modulator.

 

When we use wide band we can get audible intermodulation distortions (depend on playback  equipment and software).

I'd recommend use 20 kHz limitation (optimized filter mode in the Settings > General of AuI ConverteR). But also I have feedback by some our customers, that wide band filter 100 kHz (wideband non-optimized filter mode) have better sound.

 

 

1. Quality of any resamplers we can estimate by pure sweep sine, that modeled mathematically. Musical signal is too complex for estimation, because the music spectrum hide distortions (music is mixed with the distortions).

 

2. Spectrum analyzer should provide -180 ... 200 dB own noise floor in 150 kHz and more band without filtering aliases. Because DSD modulators at DSD256 and higher may have noise floor about -200 dB and better. Also DSD modulator can provide band up to 100 kHz and higher.

 

I don't know about such anaylizing software for 1-bit audio files. Therefore we use own custom analyzis software.


Additional information about DSD converters audio (including modulator and demodulator) here http://samplerateconverter.com/content/what-inside-dsd-converter-audio-files

AuI ConverteR 48x44 - HD audio converter/optimizer for DAC of high resolution files

ISO, DSF, DFF (1-bit/D64/128/256/512/1024), wav, flac, aiff, alac,  safe CD ripper to PCM/DSF,

Seamless Album Conversion, AIFF, WAV, FLAC, DSF metadata editor, Mac & Windows
Offline conversion save energy and nature

Link to comment
20 hours ago, klassikmann said:

None of the five resamplers show a lowpass filtering which is IMHO conformant to Sony's recommendation to apply a filter with a cut-off frequency of about 50 kHz.

(...)

2. @copy_of_a: Do you keep your verdict about XiSRC and the conversion of DSD material? If yes, why?

In addition to Yuris comments...

 

First, the foobar (both) and TEAC conversions show lowpass filtering! The source file contains a noise level short below -60db at 50kHz. It's filtered out in the foobar conversions and attentuated  in the TEAC conversion.
mansr enhanced SoX and XiSRC preserve the "DSD-noise".


Personally I don't think DSD to HiRes PCM conversions require steep (brickwall) filters at 20kHz (22kHz). But a gentle filter at around 30kHz would make sense to reduce the possibilty of intermodulation distortion caused by the high noise level above 30kHz. The TEAC conversion looks like there's a gentle 30kHz lowpass filtering to tame down the "DSD-noise" BTW...
But the XiSRC conversion from DSD64 to DSD128 (the first in my posts above) clearly doesn't make sense. The final file is DSD64 "rewrapped" into DSD128. So the high noise level above 30kHz of the original DSD64 is simply passed through. The DAC handles the filtering as with native DSD128... but in DSD128 the rise of the noise level sets in at much higher frequencies. 
Then again... to be honest it's beyond me why anyone would upsample DSD64 to a higher rate DSD ...

____________________________________________________

Mac Mini, HQPlayer | iFi Zenstream | Intona 7055-B | Singxer SDA-6 pro | Vincent SV237 | Buchardt S400 | SPL Phonitor One | Beyer DT1990pro | Avantone Planar II
Desktop: Audirvana Origin | Intona 7054 | SMSL M500MKII | NAD C 316BEE | Aperion Novus B5 Bookshelf | Lehmann Rhinelander | Beyer DT700proX

Link to comment
2 hours ago, copy_of_a said:

In addition to Yuris comments...

Thank you for your comments, which are highly appreciated. I contacted XiVero about lowpass filtering and the answered that they use an 50 kHz ultrasonic filter for DSD64 conversions, but no filtering for higher DSD resolutions because of their much lower noise floor. I'll keep in contact with XiVero, smart and friendly people.

 

To my hardware environment and main resampling scenario:
There is no dedicated DAC in place. My Raspberry PI with RuneAudio is digitally connected to the (active) Genelec 8351, which are PCM capable only. I decided for my DSD material not to use online resampling of RuneAudio (MPD based) but offline resampling to PCM(24/96) which is the most appropriate sampling rate according Genelec's support people to avoid an additional internal resampling.

 

And at last:

2 hours ago, copy_of_a said:

Then again... to be honest it's beyond me why anyone would upsample DSD64 to a higher rate DSD ...

I subscribe to this point of view.

 

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...