sacd_extract DSF output problem

[Gators]

Deal.

 

Meanwhile, I did start a new thread with the intended purpose of having a place to post any procedures that people are currently using to do this properly, for the purpose of having a place separate from this mess that is an easy one stop thread for procedures without having to wade through all this thread. Any postings there would be greatly appreciated, and hopefully helpful to many others that come by here.

 

I have so far successfully gotten a pop free conversion using jriver to convert the extraction I got from iso2dsd. If there's a simpler way to do this with less steps though I'd love to hear it...but maybe there isn't. I still don't have a way to playback dsf files without issues between the tracks, foobar seems to just be plain and simple bad, and maybe jriver needs some configuration, I'm not sure.

[kumakuma]

I have so far successfully gotten a pop free conversion using jriver to convert the extraction I got from iso2dsd. If there's a simpler way to do this with less steps though I'd love to hear it...but maybe there isn't. I still don't have a way to playback dsf files without issues between the tracks, foobar seems to just be plain and simple bad, and maybe jriver needs some configuration, I'm not sure.

 

Glad to hear that you've got this far. What kind of DAC are you using? Does it support DSD files?

[Gators]

I'm not using anything that supports DSD. My soundcard is a delta 1010lt, so anything I'm doing involves having software convert DSD to PCM, which it seems is where the trouble lies.

[kumakuma]

I'm not using anything that supports DSD. My soundcard is a delta 1010lt, so anything I'm doing involves having software convert DSD to PCM, which it seems is where the trouble lies.

 

In JRiver go:

 

Tools -> Options -> Audio -> Settings -> DSP & Output Format -> Output Format

 

Change "Greater Than 384,000 Hz" to a sample rate your card can handle

 

Make sure Output Encoding on the same page is set to "None"

 

This should cause JRiver to convert DSD files to PCM while you are playing them. JRiver can also handle ISO files directly.

[Gators]

Cool.

 

Looks like that setting automatically updated itself already.

 

This time I tested with dsf files I made using iso2dsd, and the transition was flawless. I don't know why I was getting the crossfading issue earlier. I know I saw a setting for gapless playback for songs within albums. I'm thinking.....when going between different songs, it crossfades by default, but plays straight gapless for same album tracks, and somehow earlier in my testing it wasn't playing or detecting the tracks as part of the same album? I dunno....but it works great now. It seems iso2dsd is the answer, and if you need to have a pcm version of something, using jriver to convert to flac is the way to go. I also now have a dvd-a/v made of a corrected Mahavishnu Orchestra - Birds of Fire, so I'll finally be able to hear that one as it's supposed to be heard.

[anoutsos]

Hi all,

I am new here and I have read most of this thread. I have used Weiss Saracon to convert DSD to AIFF (PCM) and I get the infamous pops at the beginning of tracks (see image). I have tried the simple method of selecting the offending spike and applying interpolation, which removes it. I am very close to eliminating this spike from all my DSD converted songs, but it may take the rest of the year to do it. Nevertheless, is there an automated/scripted way of doing this? The current procedure is "Open AIFF", "Zoom into the beginning of the track", "Selecting the spike", "Applying interpolation" and "Saving". I have 1,613 songs, so it would take a hell of a long time, but I am almost crazy enough to do it (then back up the songs thrice!). I have a Mac, by the way. Apart from the expensive software that people have suggested, is there another way of eliminating the spike?

 

Thank you all in advance.

Aris.

 

[One and a half]

Hi all,

I am new here and I have read most of this thread. I have used Weiss Saracon to convert DSD to AIFF (PCM) and I get the infamous pops at the beginning of tracks (see image). I have tried the simple method of selecting the offending spike and applying interpolation, which removes it. I am very close to eliminating this spike from all my DSD converted songs, but it may take the rest of the year to do it. Nevertheless, is there an automated/scripted way of doing this? The current procedure is "Open AIFF", "Zoom into the beginning of the track", "Selecting the spike", "Applying interpolation" and "Saving". I have 1,613 songs, so it would take a hell of a long time, but I am almost crazy enough to do it (then back up the songs thrice!). I have a Mac, by the way. Apart from the expensive software that people have suggested, is there another way of eliminating the spike?

 

Thank you all in advance.

Aris.

 

[ATTACH=CONFIG]29882[/ATTACH]

 

What software are you using to playback?

[anoutsos]

I am using Amarra (v3.0.3), in playlist mode. To iron out the spike at the beginning of the tracks I use Sound Studio. I have noticed that after eliminating the spike and saving, the AIFF file is different by a few kb. Could I be screwing up the bit-stream by messing with it like that? The interpolation is applied very locally, over a few tens of microseconds around the spike.

[audiventory]

I have noticed that after eliminating the spike and saving, the AIFF file is different by a few kb.

 

I suppose, it is metadata size difference.

[One and a half]

I am using Amarra (v3.0.3), in playlist mode. To iron out the spike at the beginning of the tracks I use Sound Studio. I have noticed that after eliminating the spike and saving, the AIFF file is different by a few kb. Could I be screwing up the bit-stream by messing with it like that? The interpolation is applied very locally, over a few tens of microseconds around the spike.

 

With a few experiments, the type of player can suppress the spike. Audirvana+ does a great job, so does JRiver (Win). HQPlayer doesn't fare so well. So is it the content or the handling? I use the bogi method which is on Windows only, an extraction to DFF, then to DSF as a batch process, not MAC so cannot compare, I don't have the tools for MAC like Saracon. A comparison of different players is at this post.

[anoutsos]

With a few experiments, the type of player can suppress the spike. Audirvana+ does a great job, so does JRiver (Win). HQPlayer doesn't fare so well. So is it the content or the handling? I use the bogi method which is on Windows only, an extraction to DFF, then to DSF as a batch process, not MAC so cannot compare, I don't have the tools for MAC like Saracon. A comparison of different players is at this post.

 

 

Oh, I see. I also used ISO2DSD, but I thought that the spike is introduced at the next step, when one converts DSD to PCM. I have read somewhere (“Pop” goes DSD? Why does this happen?) that the pop happens because of some sort of levels normalisation. However, the point remains that there is a spike in the data stream and it would be good to get rid of it at the extraction or conversion stage. Also, if A+ and other players suppress those spikes, does it mean that they are not bit-perfect or are they selectively alter only that particular feature of the data stream?

[Synfreak]

It is a known "feature" of Saracon to add this spike to almost all conversions from DSD to PCM (AIFF, WAV, FLAC ...).

Korgs Audiogate software sometimes (!) exhibits the same problem.

Haven't tested the TASCAM/TEAC software (Freeware) for this.

[audiventory]

I have read somewhere (“Pop” goes DSD? Why does this happen?)

 

Here I collected reasons why border clicks appear for ISO to DSF/PCM conversion (with pictures) How convert ISO to DSF WAV FLAC AIFF without clicks. User manual

 

that the pop happens because of some sort of levels normalisation.

 

Level normalization should be at single value for all album files.

 

First we measure peak level across all album tracks.

 

Second we increase/decrease level each track by identical value (dB).

 

Otherwise we damage level integrity of the album.

[One and a half]

Oh, I see. I also used ISO2DSD, but I thought that the spike is introduced at the next step, when one converts DSD to PCM. I have read somewhere (“Pop” goes DSD? Why does this happen?) that the pop happens because of some sort of levels normalisation. However, the point remains that there is a spike in the data stream and it would be good to get rid of it at the extraction or conversion stage. Also, if A+ and other players suppress those spikes, does it mean that they are not bit-perfect or are they selectively alter only that particular feature of the data stream?

 

Audirvana + can suppress the "flow" of data, since the main issue is to handshake with the DAC. The player can selectively null the output when the fs changes from one to the other. In addition there's a time value which can be programmed to allow sufficient time for the DAC to change sample rates. Thus the output from the player waits for the DAC, then starts playing. I don't think the app is smart enough to detect nor should it suppress the spike. As Synfreak says, the spike is a feature with Saracon

[anoutsos]

Audirvana + can suppress the "flow" of data, since the main issue is to handshake with the DAC. The player can selectively null the output when the fs changes from one to the other. In addition there's a time value which can be programmed to allow sufficient time for the DAC to change sample rates. Thus the output from the player waits for the DAC, then starts playing. I don't think the app is smart enough to detect nor should it suppress the spike. As Synfreak says, the spike is a feature with Saracon

 

So, the spike is a feature of the conversion from DSD to PCM, but there are players, like A+, that suppress it? Does the ISO to DSD extraction introduce this spike, too?

Also, if part of the data flow is suppressed by A+, how does A+ reproduce gapless recordings? Does this introduce gaps during playback?

 

I also wanted to ask about the difference between Bogi's ISO2DSD and using sacd_extract from the command line. Do they lead to exactly the same output? I am asking because I would rather script sacd_extract for batch jobs than load them into a clunky graphical JAVA interface.

 

Thanks.

[bogi]

sacd_extract 0.3.8 claims that the bug with truncated last data block was corrected. That bug was the reason I created ISO2DSF. There are many reasons why click on track transition can occur, that truncated data block is only one of them. Difference in signal levels on end of one DSD track and beginning of the next is more common reason. Switching between PCM and DSD DAC output modes is also known to produce clicks. In the case of DoP switching DSD -> PCM -> DSD occurs always between tracks. It depends on DAC implementation and SW player implementation if these clicks are avoided by playing a silence during track transition or DAC output mode switching.

 

I also wanted to ask about the difference between Bogi's ISO2DSD and using sacd_extract from the command line. Do they lead to exactly the same output? I am asking because I would rather script sacd_extract for batch jobs than load them into a clunky graphical JAVA interface.

 

My ISO2DSF works in 2 steps

1. ISO to DFF + extracting metadata via sacd_extract from Mr.Wicked

2. DFF to DSF + metadata import via dff2dsf from Signalyst

 

So it is not the same as running sacd_extract directly to DSF. What is different and could cause a difference is metadata processing.

For example I am adding SACD catalog number into id3v2 tag comment field and I have no idea if sacd_extract does the same. But regarding clicks there should be no difference between the two tools if sacd_extract 0.8.3 or newer is used.

[anoutsos]

Thank you for your reply. Does the latest ISO2DSD for MacOS X use sacd_extract v0.3.8 or v0.3.6? As for the format, I have been extracting DFFs (converted to DSD), because Saracon does not recognise DSFs. I remember that at some point it was being suggested that the solution to avoiding the clicks was to first extract as DFF and then convert to DSF. I don't think that this is valid, though, since I have been extracting directly to DFF and I still get the clicks. By the way, if I am only interested in DFFs, would I not get the same output by using the "-p -c" flags in sacd_extract? As far as ID3 tags are concerned, I typically add them later anyway. sacd_extract parses the names of the tracks and that's enough for me.

 

 

Actually, I've just compared v0.3.6 to the one supplied with ISO2DSD. The former works twice as fast! So, there must be quite some difference.

[audiventory]

sacd_extract 0.3.8 claims that the bug with truncated last data block was corrected.

 

After correct extracting DFF from ISO there is need sometimes cut pauses between tracks during converting to target format (output format of audio converter). Into these pauses may be clicks and wrong data (not for all ISOs of course).

 

Also may be issues of long time file names.

Here I meant base sacd_extract [not ISO2DSD] version 0.3.8 under Windows. May be ISO2DSD work differently.

 

ISO with long and non-English encoding track names demand additional automation under sacd_extract.

For Win version also exists other issues with long target file names, what need to solve.

 

Under Mac sacd_extract 0.3.6 work without file name issues (I don't know such issues currently).

[bogi]

I remember that at some point it was being suggested that the solution to avoiding the clicks was to first extract as DFF and then convert to DSF. I don't think that this is valid, though, since I have been extracting directly to DFF and I still get the clicks.

 

Read more carefully preceeding posts. There are more reasons why clicks may occur.

[anoutsos]

I am planning to use sox to trim 0.0006 s from the beginning and end of the audio files (where the chirps are always found). My command is

sox input output trim 0.0006 =(length-0.0006)

Can anyone see any problems with this approach?

[paul30d]

I am planning to use sox to trim 0.0006 s from the beginning and end of the audio files (where the chirps are always found). My command is

sox input output trim 0.0006 =(length-0.0006)

Can anyone see any problems with this approach?

I use ISO2DSD to create a DFF Edit Master and cue file. I then convert the DFF file to one large flac file and split the file using XLD with the cue file. With this method, the only place that a click can occur is at the beginning of the first song of the album or the end of the last song of the album. But I have never heard (or perhaps noticed) a click at the start or end of an album. I think that is due to converting first to DFF, which apparently avoids the error.

[bogi]

I am planning to use sox to trim 0.0006 s from the beginning and end of the audio files (where the chirps are always found). My command is

sox input output trim 0.0006 =(length-0.0006)

Can anyone see any problems with this approach?

 

That's interesting. You may ask mansr, he implemented DSD support in sox.

http://www.computeraudiophile.com/f11-software/direct-stream-digital-encoding-sox-25556/

[bogi]

I use ISO2DSD to create a DFF Edit Master and cue file. I then convert the DFF file to one large flac file and split the file using XLD with the cue file. With this method, the only place that a click can occur is at the beginning of the first song of the album or the end of the last song of the album. But I have never heard (or perhaps noticed) a click at the start or end of an album. I think that is due to converting first to DFF, which apparently avoids the error.

 

I think significant part of SACD ISO users does not convert to PCM. We can create 'edit master' (big DFF + CUE), but how to split it to single DSD tracks? Available CUE splitters are only for PCM format (AFAIK).

[Synfreak]

I think significant part of SACD ISO users does not convert to PCM. We can create 'edit master' (big DFF + CUE), but how to split it to single DSD tracks? Available CUE splitters are only for PCM format (AFAIK).

 

That would be a nice one, having such a "splitter"!

Because I have lots of (mostly classical) SACDs, where sacd_extract seems to produce false splits - no matter if the target file is *.dff or *.dsf.

Currently I am really thinking about to incorporate the ISOs into my music library instead of the extracted tracks because of this ...

[audiventory]

Available CUE splitters are only for PCM format (AFAIK).

 

Some CUE splitters can split DSF+CUE (see my signature) ;-)

One my customer gave me this idea that was released some time ago http://samplerateconverter.com/content/new-demo-policy-dff-output-windows-only-dsf-cue-splitter