Beta-voltaics are _weak_. Nickel-63 emits beta-particles at 66kEV, and you can recover less than 1% of that, so around 660eV per electron (a beta-particle). So if you need 200W to power the satellite, you need around 0.3 Ampere of beta particles. That's 2*10^18 particles per second. Nickel-63 has specific activity of 2*10^12 Bq/g, so you need around 10^6 grams of Ni-63, or around 1 ton.
You can get a more active isotope with a shorter half-life, like tritium. That'll reduce your weight by around 100x (tritium is around 4*10^14 Bq/g), and perhaps you can use more efficient generators. But we're still talking about at least hundreds of grams to kilograms of radioactive material. That's waaaay too expensive.
On the other hand, 200W can be easily generated by 1 square meter of solar panels.
Now, if you want an underwater microphone that might occasionally communicate with other sensors via bursts of encoded ultrasound, you need probably around 1W on average. So that immediately drops the amount of radioactive isotope from kilos to several grams. That's way more feasible.
There's evidence that the USSR was doing exactly that because it produced more tritium than required for the nuclear weapons program and for accounted civilian use. Tritium is also great because it's produced naturally, and if one of your microphones leaks, it'll be much more difficult to trace.