DirectStream Displaying Unexpected Bit Depth

I purchased a 96/24 FLAC recording of Jimmy Rogers Blue Bird but my DirectStream displays 96/20. The frequency spectrum seems to indicate a full range file and not a converted lower resolution source. Has anyone else experienced this behavior? I contacted the seller and they are looking into the display discrepancy. I have other recordings from this seller that produce the expected display. Any insight as to how I can examine the raw data would be appreciated.

Equipment: Mac Mini (OS X 10.14.3) running Roon (1.6 build 401) with a direct ethernet connection to my DirectStream DAC (Snowmass 3.0.5) via Bridge II (3.6.17).

Jim

The DS is looking at the bottom bits over a period of time, if no non-zero bits are seen in the bottom bits that’s reported. We’ve never seen it be wrong.

Still you can run the bit perfect tests to see if all bits are being transferred correctly. (See https://www.psaudio.com/ps_how/how-to-run-a-bit-perfect-test-with-directstream/)

For checking the file content: programs like Adobe Audition (and I think Audacity) have functions to check the number of bits that are being used even if a volume change has happened. I don’t know if those are available on the MAC or what software on the MAC might do that for you. Similarly you might be able to use a hex editor or displaying program to see if there’s one 0 in each set of 6 nibbles (a nibble is half a byte.)

[Edit:] I should also have said that I have some random “24”/96 files that are really 20/96, but most software reports 24/96 since 24 bits (three bytes) of storage are used per sample.

1 Like

The bit perfect test passes with flying colors. I just noticed Roon claims it’s a 96/24 FLAC file. Since I can run any OS, I’ll see what platforms the software Audacity and Audition are available on. Something for Linux would be ideal.

Do I need to uncompress the FLAC file prior to hex dumping to search for the empty nibbles? The “od” command should do the trick, but it doesn’t support a 3 byte word size.

I appreciate the rapid response to my issue. Keep up the excellent work!

Yes, you’ll need to uncompress the file. od will work fine, I’d use “od -t x1” (if your od is like Cygwin’s) The pattern of 0s will be obvious if it’s there.

Correct me if I’m wrong but conversion to AIF via XLD should perform the uncompression.

After conversion to AIFF the pattern of zeros is clearly evident. The recording was advertised and a premium charged for 96/24 which was not delivered. I’m awaiting a response from the seller.

1 Like

I wouldn’t get too upset about a 20/96 recording vs a 24/96 recording: you are talking about at most a 1 part per million difference in amplitude. It’s quite doubtful that your system as a whole can even resolve that difference. The 24 vs 20 difference is quite small compared to the 20/16 difference which can often be audible.

Don’t get me wrong you should be getting what’s advertised, but they should just, say, change their site to reveal that the source is 20 bits.

My Directstream Senior will often show a lower bit depth, like 16 bit, if I lower the volume in the music player. Make sure the volume is all the way up in Roon.

With Roon, if you have a 16-bit file playing and use any of the volume controls or DSP, what it sends out is the 24-bit result of its calculations (which are internally performed with 64-bit resolution). I would be very surprised if Roon would ever truncate 24-bit material down to 16 bits.

What player do you use?

jim-kaporch: Who was the seller?
Thanks in advance,
Jay

I agree. The site should properly document the format and price accordingly.

I have Roon set to fixed (maximum) volume, so I have never experienced this.

I am reluctant to disclose the vendor. They promptly addressed my complaint and refunded my money. They trusted the label regarding sample rate and resolution which was reflected on their website. For this recording the sample rate was correct but not the resolution. I have other recordings from this vendor and all were as advertised. They contacted the label but were unable to obtain the advertised format. I’ll check back later to see if their website has been updated. If not done in a timely manner I may reconsider disclosing the vendor.

Bottom line if purchasing digital downloads I recommend verifying format soon after purchase. It’s actually pretty simple for PCM. I haven’t a clue about DSD, I would have to research that format more.

I am using Audirvana, and pretty sure it doesn’t truncate either, and with my Mytek DAC sends out 32 bit. Seems to me like a quirk in how the Directstream recognizes input bit depth. (for one, it went straight from 24 to 16 bit with even the slightest drop in volume, when it would at least have done 20 bit first)

There is no quirk in the DS in this respect. If there are nothing but zeros in the lower bits for a while… it tells you.

Oh… are you using USB? If so, make sure you’re not using the “hardware” volume control in Audirvana. That will end up telling the XMOS USB receiver in the DS DAC to do its own volume adjustment, and I could imagine that doing a silly 16-bit truncating thing. If you’re doing volume in Audirvana with the DS via USB, make sure it’s a software volume control.

Ahhh. Thanks.