process an SACD-ISO file

[blanc]

Hi,

I want to process an SACD-ISO file so that my streamer can play it. This file includes a 2C_Audio folder, an MC_Audio folder, three TOC files and one LST file. The conventional way is to convert it to flac or wav files with Foobar2000 but according to this forum: https://forums.stevehoffman.tv/threads/convert-split-sacd-iso-file-halp.1182252/ this conversion is NOT lossless. What do you think?

It seems the best way is to get DSF or DFF files with ISO2DSD. Which tools do you recommend for Windows PC?

 

[pvanosta]

ISO2DSD is freely available for windows PC and it is what I have been using for many years now.

 

[Kal Rubinson]

Both suggestions will work fine. Saying that the foobar2000 conversion is not lossless is a semantic point in the sense that many say that any conversion from DSD to PCM is not lossless.

 

[blanc]

What about Wavpack comparing with ISO2DSD? The generated .wv files seem not so commonly used by streamers.

 

[Vincent Kars]

any conversion from DSD to PCM is not lossless.

Correct.
Inherent to DSD is the tremendous amount of quantization noise. As it is "infinite" it will exceed the Nyquist frequency of the PCM.
Even a 384 (DXD) cannot contain all the noise.
Of course one must use a low pass filter while converting to PCM. By design you can't use a filter and being lossless at the same time.

 

[blanc]

I think for conventional stereo HIFI system the MC_Audio folder is not relevant, right? Are there alternative formats that consume less space but are still lossless other than DFF or DSF.

This description is taken from that forum:

I preffer to send the dsd files directly to the DAC to not having the conversion (as principle) but also as the conversion needs to lower volume to allow some headroom and not clipping and I never settled in a value for this, -6db will render some music very low for example, -3db will make some clip…simpler to not convert.

What does it mean? Does it mean some formats will reduce the volume inherently?

 

[danadam]

A bit unrelated:
Correct me if I'm wrong, but it seems the graph is somewhat misleading in the vertical direction. The blue trace shows, I assume, spectrum of the DSD noise floor, while the bottom edge of the boxes is at -144 dB, which is integrated value for 24-bit PCM noise floor. The spectrum (or spectral density) of 24-bit PCM noise floor would be much lower than that.

I may have hit some limits in sox (mansr version, supporting dsd: https://github.com/mansr/sox ), but here's 32/88.2k PCM converted to 24/88.2k:

sox input.wav -b24 output.wav dither

and to DSD64 and to 32/2882.4k PCM:

sox input.wav -b1 output.dsf rate -v 2822400 sdm -f sdm-8
sox output.dsf -b32 output.dsf.wav

and the spectrum:

 

[10PSPMemSticksForMiku2007]

While converting DSD to normal PCM is not strictly lossless, it should sound the same in the audible range while losing only inaudible high frequency noise. DSD is such an inefficient system. I'm happy to lose that noise and not waste my drive space on it. In my collection, I convert anything with an excessive rate to 48khz 24bit FLAC and it all sounds good to me.

 

[aagstn]

When I ripped all my SACDs I kept them as DSD files, but also converted them to 88.2/24 files. The only difference I noticed was the volume was about 5db lower on the PCM converted files. I ended up just using the DSD files as space wasn't an issue for the 40 SACDs I had and I got SMSL DACs that could handle DSD. If I decided to go back to PCM I would probably boost the volume to make up for the difference.