Most of all, I would ask yourself what you want to get out of this. Is it just engineering experience, or something you'd actually make good use of (moreso than a printer you could just buy), or some combination of the two?
While more-than-3-axis printers are very cool, I would lean against that option unless you really want this to be a software project too. The big reason they haven't taken off isn't that the engineering is hard (although it does have some challenges), but that the slicing software is not there to take advantage of them, so you end up having to either do the gcode "by hand" or hack together a primitive slicer whose output is so much worse than just living with 3-axis but all the modern path generation of mainstream slicers.
I think I agree with your assessment about low-noise. Also, for high speeds, with most materials you need heavy cooling, which is a lot louder than motion system noise.
Large build volume could be interesting, especially if it's something you could use. So many of the large build volume projects I see are not really well-engineered, just scaled up with no consideration for how that affects rigidity/accuracy. Making a large form printer that prints at high quality and decent speed is a fairly hard engineering problem, and one that you might enjoy and learn a lot from.
There's also an option to try different kinematic systems, which have a lot of different engineering tradeoffs in:
- BOM (parts) size/cost/complexity
- Sensitivity to inaccuracies in build procedure
- Speed and acceleration capabilities
- Ease of enclosure
- How motion component sizes scale with desired build volume
etc. etc. etc.
Beyond whatever ideas you might get from folks here on SE, in my experience, some of the best communities for getting ideas and feedback on custom printer designs are the Rolohaun and Annex Engineering Discord servers. The former is more novice hobbyist/DIY geared while the latter is more extreme performance geared, but both are friendly to this kind of activity.