Possible FLAC/DSD Conversion Project

I am contemplating a major project, but I am looking for advice before I begin. Several years ago, I ripped my extensive CD collection (pop, rock, classical) to FLAC files. When I purchase downloads, I always purchase FLAC format, usually 44.1/16, but some are 24 bit. I listen exclusively via headphones, a FiiO M11s serves as the digital transport, and a FiiO K9Pro (AKM) is the DAC/Amp. Both support DSD up to 256. I would very much appreciate answers to the following questions:

  1. Does the difference between DSD and FLAC represent a real improvement in sound quality?

  2. Does converting (up-sampling?) a file from FLAC to DSD involve a reduction the of quality of the file? (Note: this would be my greatly preferred way, as it would preserve all of the corrections/enhancements I have made to the metadata tags in the ripped FLAC files over the last couple of years.)

  3. Considering only sound quality, would I be better served by re-ripping the CDs directly to DSD?

  4. How much larger are DSD files? (I know it would be an estimate but if a FLAC file is 100 MB, how large would a DSD file of the same CD track likely be?)

  5. What would a good ripper/conversion tool be?

  6. Finally, in your opinions, is the juice worth the squeeze? I do have the time, but I dislike wasted effort if there will not be a perceptible sonic improvement.

Thanks for reading and responding!

1 Like

As you read the following, remember that FLAC is PCM data, hence all the PCM references.

Not if youā€™re converting from one to another.

You might get a benefit if your DAC does a better job with DSD vs. PCM content, but thatā€™s DAC specific and as likely to go one way as the other.

DSDā€™s benefits come from a) allowing the use of a 1-bit ADC (which, being a single bit, has perfect linearity), and a 1-bit DAC (also with perfect linearity). If the initial capture isnā€™t done as a native DSD stream, the first benefit is lost forever. If the replay isnā€™t done as a native 1-bit stream, the second benefit is lost (most common DSD chipsets are hybrid bitstream/multi-bit converters).

If any conversion to PCM occurs between capture and replay, which it invariably does, and has already happened for the content youā€™d be converting, thereā€™s no gain beyond, again, your DAC potentially being better with DSD than PCm.

It should be audibly transparent, but it is not bit-perfect.

There is no mathematically lossless conversion from FLAC to DSD, nor DSD to FLAC. So, you canā€™t convert FLAC to DSD to FLAC and get the same bits back. Same is true with DSD to FLAC to DSD.

Wonā€™t make any difference at all; the PCM data that comes of the CD will be the same either way.

Depends on what DSD bit-rate you convert/upsample them too.

DSD64 (1x) is roughly equivalent to a 24-bit 88.2 kHz PCM file. Thatā€™d be ~3x the size of the equivalent raw PCM for a CD, and about 6x the size of a typical FLAC copy of it. So, a 100MB track in 16/44.1 kHz will be about 600MB in DSD64.

DSD128 (2x) doubles that (1.2GB), DSD256 (4x) doubles it again (2.4GB) and, of course, DSD512 (8x) doubles that again (4.8GB).

dbPowerAmp is a nice tool for doing CD rips, and I believe it supports conversion to DSD (it canā€™t rip FROM DSD). No idea how good its DSD conversion is.

The ā€œgold standardā€ for converting PCM to DSD is HQPlayer. Unfortunately, you have to have the ā€œProā€ version if you want to output the results to a FILE. From memory, the ā€œProā€ version is $3-4,000 (but check). The ā€œDesktopā€ version is ~$240, but is a PLAYER and doesnā€™t save its output to files.

Unless your DAC is obviously better at DSD replay than PCM, thereā€™s no benefit to it.

The only exception that really comes to mind is that sometimes using HQPlayer to do extreme DSD upsampling and filtering yields better results that whatā€™s built into your DAC.

Iā€™ve not heard any AKM based DACs that did better with DSD than PCM (doesnā€™t mean there arenā€™t any, Iā€™ve just not come across them).

11 Likes

Wowā€¦ thank you for a most complete set of answers to my questions! I did suspect as much, but I wanted to get opinions from ones more knowledgeable than I. Thanks for taking the time, Torq.

I can answer this in fewer wordsā€¦ you wrote that you are using the FIIO M11sā€¦ it has a DSD DAC that has a setting where it converts to DSD on the fly while playbackā€¦ just turn that option on and if it sounds better for you then use it ā€¦ if you donā€™t like it then turn it offā€¦ pretty simpleā€¦ and there are Music Apps that perfom simular magic like the Onkyo Hifi Playerā€¦ and its a lot easier than you converting any amount of songs you haveā€¦ give it a try and let us know how easy it is and if you like it or notā€¦ should take about 5 seconds

1 Like

In my experience the surface flaws and defects of physical CDs are much, much more noticeable than any digital format above 16/44.1. A couple years ago I compared my lossless CD rips to the same files from online streaming services. In most cases I preferred the streaming versions. First, some of the services have stealth remastering at higher bitrates so itā€™s hard to directly compare ā€“ but they were surely more nuanced than a lot of 1980s-1990s early CD mastering. Second, many of my home rips had obvious pops, skips, and other defects printed to the discs.

I used Exact Audio Copy and compared my rips to their online databases, and had a 5% to 10% disc flaw rate (any errors per disc) across hundreds of CDs. I also confirmed that USB cables must be <2 meters for reliable ripping with an external drive.

Test with Exact Audio Copy for free. Itā€™s clunky and dated but itā€™ll get everything there is to get from a CD, and it references several online disc databases to check for accuracy.

No. No. No. No. No. Try an uncompressed and often higher bitrate streaming service for a month for yourself (Qobuz, Tidal, Amazon HD, etc.). Many original sources were poorly recorded and cheaply produced too, so you may end up focusing more on the flaws than the music anyway.

2 Likes

Thanks for the suggestion. I did notice that, and tried it. Two things happened. First, the battery was exhausted VERY quickly, lasting not even two hours. This isnā€™t so much of a concern since it would be plugged in all the time. I am not using this as a mobile device, more of a ā€˜transportā€™ to feed the desktop DAC/Amp. Second, the M11s got blisteringly hotā€¦ really, really hot. Heat is usually the enemy of electronics so this worries me long term, since I would be doing this for every piece of music I listen to.

Iā€™ve been using the M11s this week and while the ā€œAll to DSDā€ did use a lot of battery, I didnā€™t really notice much heat increase (although Iā€™ve seen it mentioned a lot).

I used/use dbPoweramp as my ripping tool, and it has Accu-rip built in. I have it set to always check the ripped file, and have had very few (perhaps five) discs/rips that did not pass the ā€˜testā€™. Iā€™ll run a few through
EAC and see if I get different results.

I may sign up for a trial of a streaming service and try a comparison of one of my FLAC files with a streamed hi-res version of the same file.

I only did it with a couple of symphonies, perhaps two hours total, and mine got VERY hot. Perhaps mine is faulty in some wayā€¦ I hope not, itā€™s only a couple of weeks old!

Error checking routines vary. I actually re-ripped my collection because back in the day Appleā€™s iTunes/Music app said there were no errors when there were many and major errors. Still, even with a validated rip I mainly use streamed files now.

1 Like