So the other answer which delves into the calculation side is very good, and use that for a theoretical basis for your design. I'm not very good at theory, so I experiment instead.
I'm currently designing a printbed for a commercial 3D-printer that is supposed to be able to cool the print down to 0 degrees C. In practice this means that the surface of the printbed will need to reach sub-zero temperatures.
I can tell you straight away that for a block of tofu that large, if you only cool from a single direction (i.e. from the bottom as you do with a printbed for example), your main problem will be that you will be able to cool to bottom part of the tofu to sub-zero without any larger issues, but the top part will not become very cold. I can get my printbed surface down to < -10 degrees C without much issues, and then I'm feeding ~120W into it, with a good heatsnik and a 10W fan on the bottom side of the Peltier element. But since the object I'm trying to cool is exposed to the ambient surrounding, the top part isn't affected much by the cooling.
If you really want to coold the entire block of it you will need to cool from several sides. I've done Peltier cooling with two blocks that squeeze around the object to be cooled, but for your relatively large volume you might even need to cool from even more directions, or build a completely enclosed volume to cool it in (like a refridgerator).
When it comes to component selection, don't underestimate the value of a really good heatsink. Heatsinks, in my experience matter more than fans, increased airflow across the heatsink has very rapidly diminishing returns. If you want to increase the performance even more, go for liquid cooling. We did an experimental build with PC liquid cooling on our Peltier elements, and the performance gain was quite massive over air cooling, but you will always have some leakage somewhere with liquid cooling.
For the Peltier elements themselves, due to construction & physics, Peltier elements designed to run at ~12V seem to be slightly more efficient than elements designed for 24V. I'm not entirely certain why, an expert explained it to me but I didn't fully grasp it, more than that "this is just the way it is, due to how they are designed & built".
For electrical components, if you want to get started really quickly with experimenting, get a somewhat beefy H-bridge evaluation board that you can feed PWM control inputs to. Hook it up to an Arduino and a power supply, and then start to experiment. Also, never run a Peltier element without cooling it, if none of the sides are cooled it will VERY quickly self heat and destroy itself. Peltiers are very fragile.
EDIT: Also, remember to use a thermal paste that is suitable for sub-zero temperatures. We use Thermal Grizzly Kryonaut for example.