I listen to classical infrequently, but I am a bit partial to Chopin, as my father practiced it a lot during my childhood. I never got past a fair rendition of the Moonlight Sonata (simplified?).
Anyway, there was a short review of this site of Tippett Rise Studio on Audiophile Style. They have a fair amount of DXD files for free download.
I learned that DXD is not a variant of DSD, but rather (from another site):
24bit 352.8kHz PCM Flac (DXD) - A digital recording at 352.8 kHz sounds closer to “real analog,” which needs no anti-aliasing filters. The listener can better capture the full ambient information of sonic events with a 352.8 kHz sampling rate. Even though a higher resolution and higher sampling rates sound better than a lower resolution and lower sampling rates, 192 kHz, 96 kHz, and 44.1 kHz/24 bit all sound more “digital” than 352.8 kHz sampling rates because of the pronounced effects of anti-aliasing filters in the lower sampling rate systems. DXD is only playable on compatible hardware and software.
I’m currently running Antipodes Oladra to Ref DD via HDMI I2S, I downloaded the DXD files and they are 32/384. However the display on DAC shows 24/384. Does it mean the I2S module only supports up to 24bits instead of 32bits? I can confirm Oladra supports 32bits I2S output.
I am using ProUSB and Antipodes K21 streamer to the Cascade DAC.
When i played 32/352 kHz files, the Cascade DAC displayed 24/352 kHz.
When the file is 24/352 kHz, the display on the DAC shows 16/352 kHz.
Pro-USB is limited to 24/352 due to software / hardware glitches according to Dustin in another post and I realized the Universal I2S has the same limitation. That said, when I switched to the Renderer (direct stream from Oladra) and played the same file (32bits floating), it still shows 24/352.. that puzzles me as the website clearly states it support 32bits. Maybe MSB process the file in its best possible way?
First off 32bit floating point does not contain 32bits of audio DATA. It is essentially 24bits (or less) of audio with space for 8 bits of nondestructive volume control information (for editing masters nondestructively). It is a convenient format internal to an editing system but not really intended to be used for distribution of music. The 8 bits instruct the software to move the decimal point to different positions inside the 24bits of data. It is up to the player software to translate that to fixed point for playback, which is the only format physical DACs can deal with (physical DACs of any current design do not deal with decimal points in their hardware) MSB DACs use a maximum of 32bit fixed point (which is the highest resolution format for PCM digital that is “standard”. A lot of software (at least in the past) has problems if it is given the option to playback at anything more than 24bits. We chose to make the USB implementation only work to 24bit because it makes the majority of software the happiest, and it is the maximum that any real ADC will output anyway. If the 32bit float files really are unedited then they will always be translated to 24bit fixed point (or less) anyway. The only time you would get more than 24bits from a 32bit floating point file is if there were volume edits made after it was recorded. Even then the playback software will probably choose to just output 24bit fixed point because that format has the widest compatibility and avoids truncation by DACs that cant handle 32bits.
The ProISL system also supports 32bit (fixed point), its just the USB implementation that supports 24bit maximum for good software compatability.
One more note, files that are 16 bit or 24 bit or 32 bit are just what the file has space for. A lot of times a 24 bit file may only contain 20 bits of information. The DSP in the DAC literally counts the used bits. So what is displayed on the DAC is a real representation of the actual bits used for playback not what the file specifies (which could be a greater bit depth than the actual audio data it contains).
