"Real time" audio classification #27
Description
The current audio classification depends on a recorded audio transmission. This results in a demodulator being tied up until classification occurs and only then possibly rejected as not interesting to the user.
This change would move classification into audio stream as it is being written to disk. Classification currently only uses two seconds of audio so this can be done much earlier in the demodulation.