Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ted_b]

Hi.  I am a go-to (not to be confused with expert) person on Computer Audiophile about things like SACD ripping, DSD, etc.  It seems some folks' systems exacerbate the tiny clicks between tracks that the current sacd_extract v37 (or below, frankly) "adds" to any SACD ISO extraction when choosing DSF as a format.  It doesn't bother me and my system, but it's clearly there on some (not a FLAC vs WAV thing at all; i.e not debatable).  And DSF is the favored format mainly due to ID3V2 tagging support.  So...there is the Audiogate solution, which says extract to DFF (no clicks) then convert to DSF using Audiogate (and it's bothersome auto-twitter requirements).  Once there, we can use any number of solutions for tagging, such as "fill propertiies from filename", paste tagging if one has a redbook or other version of the album, or even a decent paste tag from the ISO (load ISO into MC and paste tag at least the basic fields).

Enter MC's recent DSD convert function.  Why not just stay in JRiver and use it's DSF conversion to go DFF to DSF.  It works!  But.....it has been speculated (over on the Mac forum) by user 6233638 that the conversion to DSF requires a step into PCM, then back to DSF (hell, maybe Audiogate does this, but I didn't think so).  I have to admit I had not yet seriously listened to any of the test albums I've tried this on (cuz I really didn't care; I don't need to do this step in my daily routine) to determine if the result is any different than Audiogate's.  Technically, AG added 3kb to the file size (114,142kb vs 114, 145kb), likely header differences; JRiver had a paltry 4kb more than AG (i.e 7kb), with no changes to what is captured to the convoluted track name tag...i.e seems there is same going on here.  But.....when I converted the DFF file to DXD (24/352 wav, a hirez resolving PCM sample rate used in DSD editing in such pro environments as Pyramix 8.0 or below) and then to DSF via JRiver the file size remains the same as the JRiver DFF to DSF.  So that opens the door to a possible PCM step.....but maybe AG is doing the same?  I could not replicate this on AG cuz it won't accept a sample rate of DXD; I tried downsampling to 24/192k then did DSF on AG and got another 28kb in file size, so this test is somewhat moot.

Net/net, is DFF to DSF on JRIver going through DXD first?  Seems like it.  If so, are the sonics affected?  To my ears BOTH DSF conversions lose a percent or two of air compared to the unconverted DFF.  I then re-extracted the ISO as DSF and it sounded identical to DFF (and I heard the click on my Mytek-based home office headphones, but it's not there on my Meitner-based big rig).  Argh.

Maybe the best solution for those click-sensitive systems is to simply use DFF and paste tag from the ISO (temporarily imported via file-open media file) or just listen from the darn ISO. 

[6233638]

It seems some folks' systems exacerbate the tiny clicks between tracks that the current sacd_extract v37 (or below, frankly) "adds" to any SACD ISO extraction when choosing DSF as a format.

Argh, I thought this had been fixed. I do not want to go through the process of re-ripping SACDs...
Is this only when converting to DSF, or extracting DSF tracks?

It doesn't bother me and my system, but it's clearly there on some (not a FLAC vs WAV thing at all; i.e not debatable).  And DSF is the favored format mainly due to ID3V2 tagging support.  So...there is the Audiogate solution, which says extract to DFF (no clicks) then convert to DSF using Audiogate (and it's bothersome auto-twitter requirements).  Once there, we can use any number of solutions for tagging, such as "fill propertiies from filename", paste tagging if one has a redbook or other version of the album, or even a decent paste tag from the ISO (load ISO into MC and paste tag at least the basic fields).

Do we know that Audiogate stays in DSD when converting to DSF?

Enter MC's recent DSD convert function.  Why not just stay in JRiver and use it's DSF conversion to go DFF to DSF.  It works!  But.....it has been speculated (over on the Mac forum) by user 6233638 that the conversion to DSF requires a step into PCM, then back to DSF

http://yabb.jriver.com/interact/index.php?topic=79650.msg542316#msg542316

If so, are the sonics affected?

When you convert from DSD to PCM, you have to filter out all the ultrasonic noise. (by default, Media Center uses a 24kHz 48dB/octave low pass filter)
When you convert from PCM to DSD, you are adding a lot of high frequency noise into the file.
This is not going to be a transparent process.

or just listen from the darn ISO

I had avoided this because dealing with ISO files is a pain, and this was prior to Media Center having the option to convert SACD ISO to individual tracks. (if I wanted to convert to ALAC for a portable device, for example)
 
Perhaps that is the best solution for now.

[ted_b]

6233638,
Your link to support your stance that native tracks converted from DFF to DSF tracks is through PCM is not pertinent.  Matt was simply stating that in order to convert an ISO image (one large file) in JRiver one needs to edit and split the tracks.  That clearly (in JRIver) is a PCM function.  No debate.  Nowhere in your link does it say that converting already split DFF tracks to DSF tracks is a lossy function via PCM.  I'm not arguing with you (my tests support a possible PCM step) but I am not using that link as any defense either. 

Mr Wicked's v37 is the latest extractor, released Dec 2012.  It continues to have the weird anomaly that adds a click (inaudible or barely audible in many highly resolving systems, like mine).  This is not present when extracting to DFF.  And folks claim it continues to have no clicks when going from that extracted DFF to DSF via Audiogate.  Hence my thread post.

[Hendrik]

Any conversion in MC has to go through PCM at this point, the architecture doesn't support anything else right now.

[ted_b]

Any conversion in MC has to go through PCM at this point, the architecture doesn't support anything else right now.

Thanks!

[ST]

..When you convert from PCM to DSD, you are adding a lot of high frequency noise into the file.

This is the part I am bit confused. If you send DFF files to Mytek, do you think filtering still takes place in JRiver domain or Mytek takes over and filter according to your filter, i.e. 50 to 70kHz?

[6233638]

This is the part I am bit confused. If you send DFF files to Mytek, do you think filtering still takes place in JRiver domain or Mytek takes over and filter according to your filter, i.e. 50 to 70kHz?

If you are bitstreaming, the file will be sent to the DAC as-is, and the DAC will be what filters out the ultrasonic noise.
If you use DSD Encoding, the file has to be converted from DSD to PCM (which must be filtered) and then from PCM back to DSD. (which adds noise that needs to be filtered out by the DAC)
If you output PCM from Media Center, it will handle the filtering and the DAC shouldn't do anything. (as it's being sent PCM and not DSD)

[AndrewFG]

Any conversion in MC has to go through PCM at this point, the architecture doesn't support anything else right now.

Hendrik, Just as a matter of interest does the conversion really go via PCM or does it (resp. could it) go via DoPE encoding? I am not an expert on the latter, but I had somehow assumed that DoPE is just a mathematical bit manipulation and not a time/frequency domain conversion requiring low pass filtering and stuff. Or ??

Edit:  I am specifically referring to the OP in which he referred to a "conversion" from DSD DSF to DSD DFF. Surely that could go via DoPE with no problem ??

[ST]

If you are bitstreaming, the file will be sent to the DAC as-is, and the DAC will be what filters out the ultrasonic noise.
If you use DSD Encoding, the file has to be converted from DSD to PCM (which must be filtered) and then from PCM back to DSD. (which adds noise that needs to be filtered out by the DAC)
If you output PCM from Media Center, it will handle the filtering and the DAC shouldn't do anything. (as it's being sent PCM and not DSD)

Thanks for the reply 6233638. This is from Wiki JRiver:_

Bitstreaming

Media Center can optionally bitstream the following formats:
## Dolby Digital (AC3)
## DTS
## Dolby TrueHD
## DTS-HD

Only consider bitstreaming if ALL of the following are true:
1. You have more than two speakers connected
2. You are using an HDMI or S/PDIF connection
3. You have a receiver (or outboard decoder) capable of decoding formats like Dolby Digital and DTS (and preferably the latest formats like Dolby TrueHD, etc.)
4. You have properly configured the receiver levels, speakers, distances, etc.
5. You do not want to use VideoClock, Room Correction, or any other audio features offered by Media Center

Select 'Bitstreaming' as the connection type to use an external decoder when possible.
*******************

I think the DSD too was there in selection for the newer version. Now, I want Mytek to do all decoding including using its internal clock and JRiver to send the data only. That is the best recommended setting for the Mytek. Mytek is set to 2 x DSD and the display would show Internal Sync even for 44.1 PCM. To my ears this looks like best setting. From your suggestion it looks like you are suggesting not to use bitstreaming for PCM for the reasons you have already stated. However, my understanding is PCM is send to the Mytek and the DSD DAC process them without and further conversion. Am I wrong for the assumption?

My system is MC19 (the latest Version as of two weeks ago. Mytek FW is 1.7.6 via USB2.0.

Thank you.

[HiFiTubes]

On a related note, has anyone looked at DFF tagging support in MC lately?

I thought DFF only supported Album/Title. If I change Album title, and watch tags save, then update library from tags, it does not revert...

I thought files created with sacdextract will have an ID3 tag added but these can't be modified by MC?

[jacobacci]

Looking at the speed of the DFF -> DSF conversion of MC19 and Audiogate would suggest that
- MC19 goes via PCM (conversion is slow, suggesting some kind of processing)
- Audiogate only repackages losslessly (conversion is very fast)
Only a guess.