First step would be up-sampling by factor of 64, 44.1KHz to 2.882MHz with a good enough poly phase FIR filter.Thank you for posting this. Whereas until 2008 Sony VAIO DSD Direct Player was in theory supposed to enable users to play CDs through the 'better' DSD pathway in the Sigmatel and Realtek codec chips, another factor was subjective: CDs are usually listened to casually, as background music, but by using a CPU at full utilisation (dual core 2GHz) one would be forced (cannot use CPU for anything else) to sit down and actually listen to the sound without having to convert wav to dsf and use up storage space. The act of treating listening seriously and believing that DSD upscaling was beneficial alone would have made the CDs sound better. Back then users would be comparing CDs played back on Windows Media Player and DSD Direct Player, so the better choice is apparent. I still really like the DSD Direct Player UI, and I would like to recreate it. There is a lot of DSD related open source code which I hope will enable me to develop my own DSD Direct Player. It's not clear to me why another player would be needed when there is HQPlayer and multiple non-real-time conversion options. But the process should enable me to further deepen my understanding of software and DSD audio.
To make the filter design more practical you probably want to do that in a few stages.
Next step I did not really thing about yet but could be 1 bit quantizer with a error feedback loop.