Print

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

[uirio]

Hi everybody,
first of all sorry for my bad english.

I would like to have your suggestions about the best way of storing SACD individual tracks, if DFF or DSF. I know this is an old question but I would like to collect here all the best practice from those of you who deal with a lot of DSD files inside MC.
I'm an MC user since release 17 (now I'm at 26.0.73).

Currently I have thousands of SACD .ISO images that plays perfectly through MC (bitstreaming them to a Mytek Brooklyn DAC+ and to an Emm Labs DCC2 via a custom USB interface) but MC's SACD .ISO support is very disappointing (can't rename/move files, can't programmatically change the filename name/path fields, "album.iso;x" become "album.iso", metadata reset to data on the disc, filename path automatically change to "[Artist] - [Album]", ...) and it seems that they are not interested at all in fixing it.

Therefore I'm planning to extract DSD audio for .ISO images (and then manually update the fields in the MC database to point to the new file/format).

Before doing so, and being aware that I cannot retain the .ISOs (because they currently use 7,13TB of disk space), I would like to be sure which format to use and how the conversion should be achieved.

So the questions are:

1) which format should I choose between DFF and DSD ? and, if DFF, should I convert DST to DFF ?

I read a lot about this and it seems that the best format depends mainly on the hardware you use to play the files i.e. they should play identically but different DACs (or USB interfaces to DACs) have different behaviours playing the two formats.

I did a test with few disks (one was Pink Floyd TDSOTM); I extracted DFF and DSD with sacd_extract 0.38 (so NO -z, --dsf-nopad option), no DST to DSD conversion, and both play the same on my audio systems (specifically there are no clicks or glitchs nor silence between tracks). In any case, a test over few disks can never be exhaustive for a large collection ...

Based on what I read, converting DST to DSD only reduces the CPU overhead when playing the file; is it correct ?

And more, DSF can handle metadata like FLAC does, though MC's sidecars do their job and I'm not a fan of changing an audio file for storing metadata ...

What do you think about these two formats and how you tag them ? Can you tell about your experience using these two formats ?

2) which software should I use to extract DFF or DSF from the ISO images ?
Should or should not this extraction be lossless i.e. perfectly reversible ? I'm a little bit concerned here about the different results with different versions of sacd_extract.

As I said, I used sacd_extract 0.38 for my tests, and I binary compared the extracted stereo files with the ones extracted with sacd_extract 0.36 and with RD James's sacd_extract (say the 0.40, posted here: https://yabb.jriver.com/interact/index.php/topic,119883.msg828849.html#msg828849, NO -z, --dsf-nopad option): the DFFs are almost the same (but not identical) while the DSFs are quite different; they all play the same on my audio systems.
Do you have any hints/tips here ?

Thanks everybody for reading this long post and for every comments you would like to post !

P.S. I will also try posting a request about fixing MC's SACD .ISO support since I really think using the ISOs would be the best way ...

[Awesome Donkey]

In my opinion because DSF supports metadata and DFF doesn't, it's what should be used if you care about that kind of thing. I've used sacd_extract (the setmind fork of it actually with a DSF "glitch" issue fixed) to extract DSF files from SACD ISOs and it's turned out pretty well.

Not sure about the SACD ISO issues though. I thought the Rename, Move & Copy Files tool could rename them and move them.

[lepa]

I don't know if it is a best way but I'm using sacd_extract ==> dsf and then use BatchEncoder to convert dsf files to wavpack which will compress much better

[kr4]

I don't know if it is a best way but I'm using sacd_extract ==> dsf and then use BatchEncoder to convert dsf files to wavpack which will compress much better

Which sacd_extract?

[lepa]

Which sacd_extract?

I think it is latest setmind fork release (0.39) which I have compiled

[kr4]

I think it is latest setmind fork release (0.39) which I have compiled

Good.

[uirio]

Thank for your opinions; then it seems that DSF wins, and sacd_extract 0.39 (the setmind fork) is the software to use.

I don't know if it is a best way but I'm using sacd_extract ==> dsf and then use BatchEncoder to convert dsf files to wavpack which will compress much better

Thank you lepa for the WavPack tip: I never used it but I think I would if I will ever go the DSF way (especially for the multichannel tracks).

Just in case you're interested, the main reason why I'm planning to leave ISOs is described here: SACD .ISO tracks cannot be found anymore (yet another SACD .ISO problem)

[uirio]

Please, please fix the current SACD .ISO support at an, at least, stable point; or remove it at all.
I'm a long time MC user, I have plenty of SACD ISO images and I use MC for those and for FLAC.
I think MC is one of (if not the) best player for DSD and PCM audio (at least on Windows), and one of its key features is the tagging (and querying) ability.
Limit the support, remove it at all but please do something: supporting .ISO files this way is a nonsense; things like this (SACD .ISO tracks cannot be found anymore) should never happen, in my opinion.

[Awesome Donkey]

Honestly, I'd just recommend extracting the tracks regardless what happens, because at least you'll know they'll work and you won't have to waste time messing with ISOs. I also find having SACD ISOs a bit of a waste of space since in my use case I don't need the multichannel tracks.

I'm thinking out loud here, but the imported ISO itself might be the issue here; ISOs can be discs images of anything whether it be data, music, video, etc. If I had to guess, the "data type" for the imported SACD ISOs is set incorrectly and you might need to make sure it's set to music. For the life of me right now I'm forgetting what the procedure is to change that.

[The Dutchman]

I think the easiest way to convert dsd .iso is with https://www.sonore.us/iso2dsd.html (you have to install java)
I've done all my isos with this, disable dst (dst makes file smaller like wav to flac)

[Awesome Donkey]

There's a GUI for sacd_extract. Probably better to use that with sacd_extract if you're going to install Java anyways.

https://github.com/setmind/SACDExtractGUI

Honestly though it works perfectly/faster when doing it via the command line.

[uirio]

Honestly, I'd just recommend extracting the tracks regardless what happens, because at least you'll know they'll work and you won't have to waste time messing with ISOs. I also find having SACD ISOs a bit of a waste of space since in my use case I don't need the multichannel tracks.

Thank you Awesome Donkey, that's exactly what I'm going to do!

I'm thinking out loud here, but the imported ISO itself might be the issue here; ISOs can be discs images of anything whether it be data, music, video, etc. If I had to guess, the "data type" for the imported SACD ISOs is set incorrectly and you might need to make sure it's set to music. For the life of me right now I'm forgetting what the procedure is to change that.

I'm blind here... no idea about what you're saying... "file type" is "sacd" in tag panel but I don't know what it means...
Anyway: I'm going with sacd_extract (setmind) + DSF + wavpack, I'm preparing a script for the whole library 
Then I think I will need some work about updating the MC database to point to the new files without having to import (and tag again) them all

[pluto]

I've used sacd_extract (the setmind fork of it actually with a DSF "glitch" issue fixed) to extract DSF files from SACD ISOs

Are you sure that this fix for the DSF glitch works? I've compiled my own copy of sacd_extract from 'setmind' and I can't say that the inter-track clicks are any different to the original 'Mr Wicked' code. The ISO itself is as clean as a whistle.

While I know little about the detail of compiling source code under Ubuntu, is it safe to assume that, if I did something wrong, it could not merely cause one 'feature' to fail but I would end up with a complete mess that would not work at all? The '-z' option certainly appears on the options page and the GUI is issuing '-z' on the command line.

Any thoughts?

[Awesome Donkey]

Are you sure that this fix for the DSF glitch works?

It works for me, I've tested multiple albums that segue into each other (e.g. a couple Pink Floyd albums) and they're fine.

[pluto]

OK - assuming it's for Windows, would you mind sending me your binary to try?