New Directstream question

Not sure I am understanding the question. Sorry, maybe a bit jet lagged (in New Zealand right now).

Thanks Paul. I’m only connecting the Directstream to my Mac to log some burn-in hours. The DAC will ultimately be hooked into my audio system receiving a bit perfect signal via Roon. So all is good.

Paul: Under settings in the BitPerfect software, integer mode is a option (most audiophiles believe this setting provides the best sound), and BitPerfect will only provide this option if the DAC, allows this setting (which appears as an option on my MacMini running BitPerfect software). I’m surprised this hasn’t come up before, as it seems like something that could (potentially) provide the best results. However, as stated above, it completely fails to work in “integer mode”.

Paul:

Under settings in the BitPerfect software, “integer mode” is an optional setting (most audiophiles believe this setting provides the best sound), and BitPerfect will only provide this option if the DAC connected allows this setting (which appears as an option on my MacMini running BitPerfect software). I’m surprised this hasn’t come up before, as it seems like something that could (potentially) provide the best results. However, as stated previously, the DS (running Windom) completely fails to work in “integer mode”.

Thanks,

Chris

Integer mode vs floating point mode is a bit like “native” DSD vs DoP. Depending on the situation, there’s a potential for one or the other to induce less electrical noise in the DAC in ways that may be audible.

24-bit PCM samples are usually expected by DAC hardware to be represented in twos-complement binary integer form. The USB Audio Class specification is a bit more flexible, and like many operating systems internal programming APIs it can handle audio being represented in 32-bit floating-point format. See https://usb.org/sites/default/files/frmts10.pdf and search for “IEEE_FLOAT”. Note that the single-precision float has exactly 24 bits of resolution, so you can still have bit-perfect conveyance of PCM data inside this format. (And 16 bits of DSD can be packaged inside a 24-bit PCM sample using DoP, which is actually how most DSD gets played over USB!)

Most operating systems handle audio in floats internally. Most USB audio devices expect data to be received in float format, which gets re-formatted into twos-complement integer encoding before being passed onto the next component in the DAC.

For those USB audio devices which support receiving integers instead of floats, the USB receiver can skip the re-formatting step and just relay each 24 bit sample as it arrives. Some people find that this changes the electrical noise profile on the USB receiver in a way that improves the sound quality.

I’ve never checked to see whether the XMOS USB receiver in the DirectStream DAC supports integer mode. I would not be surprised nor at all concerned if it doesn’t.

1 Like

No need to check, it doesn’t. Thanks for the explanation. It helps to understand, when one doesn’t get the expected results during set up experimentations.

Having used Bit Perfect on my Mac Mini I concur, Paul is spot on.