Media Center 18.0.181 adds several improvements to audio bitdepth handling.
Improvements1) When using WASAPI, our recommended mode for USB DACs that don't have ASIO drivers, there is no longer a need to configure the bitdepth. It will auto-configure, just like ASIO and Core Audio. It is also possible to manually select a bitdepth in Options > Audio > WASAPI - Event Style > Output mode settings..., although this should rarely be needed.
2) Dither is applied to all integer bitdepth down conversions.
Dither provides a slightly higher quality last bit of data when there are values that fall in-between two discrete integer values. Since dither is bit-perfect unless there are in-between values (meaning you're using DSP, internal volume, etc.), it's perfectly transparent for bit-perfect playback, while providing slightly better quality in cases when DSP is engaged. Some
more on this here. The separate dither option for ASIO was removed, since dither is part of the framework now.
3) Bitdepth selection is controlled by the output plugin. It is no longer selected in DSP Studio > Output Format.
ASIO has always ignored the bitdepth selection in DSP Studio since the hardware determines the bitdepth. Core Audio works the same way and always takes 32-bit floating point. So the setting was already ignored in many cases. Also, bitdepth is the very last thing in the audio chain, so it's misleading to make it something set in DSP Studio > Output Format (which is otherwise the very first thing in the DSP chain).
4) All bitdepth handling code has merged into a single, well-optimized system. Previously ASIO had custom bitdepth code to support less common formats like 'big endian 32-bit that only uses 18-bits'. WASAPI - Event Style had custom code for '24-bit in a 32-bit container'. Core Audio had custom code for the 32-bit float handling. Now the core audio engine can provide the output plugins any of these formats.
5) Revision of DoP (DSD bitstreaming) marker strategy to make it (virtually) impossible for the hardware to lose DSD-lock, even during a data shortfall.
6) Optmizations to the bitdepth conversion system, resulting in a faster audio engine and less memory movement in certain cases.
TestingSince these changes are far reaching, and at the core of the audio engine, extra testing of the first few builds that use it would be appreciated. Make sure all your devices and output modes work as expected. Start with your volume knob low the first time you use this build. We don't have any ASIO devices that make requests in less common formats like big-endian, etc. so this hasn't been tested.
More informationThere's more information on audio bitdepth available in the Wiki:
http://wiki.jriver.com/index.php/Audio_Bitdepth