DHT22 temperature-humidity

It seems that the transmission protocol is fairly complicated, with only a single tri-state line used for bidirectional data transmission and tight timing requirements, see the datasheet here and DHT::read() here.
The protocol involves an initial operation from the host to signal start of transmission, then the actual data transmission involves variable-width pulses sent from the device to the host, with a low pulse of 50us followed by a high pulse whose length determines whether a given bit is 0 (26us) or 1 (70us). This should be doable through the Bela digital inputs. These are sampled at 44.1 kHz (i.e.: have a period of 22us), so the 50us low signal would be low for 2 samples, then the 26us high pulse would be 1 or 2 samples long, while the 70us high pulse will be 3 samples long. I think it could work.

Another alternative is to connect this to an SPI input line in parallel with a GPIO. The GPIO would write to the output to signal start of transmission, then put itself in input mode and start clocking a high-frequency clock. The incoming data would be independent from the clock, but this would act as a high-frequency acquisition device which would definitely give you enough resolution. You won't need this, the above solution should be just fine.