What algorithms are available for optimizing estimated bit clock recovery of an audio FSK modulated signal over an extremely low S/N ratio channel? (e.g. barely audible due to high ambient noise and low max available signal power)
Assume the audio signal is very long in duration and the payload is suitably pseudo-randomly pre-scrambled. Even if the data cannot be recovered (or perhaps without massive super-redundant FEC), can the clock be recovered accurately?