A new design of blind equalizer is proposed that can decode a magnetic
binary channel whose bit-rate is unknown. It is shown that the veloci
ty can be estimated from information already known from a fixed delay
tree search (FDTS) decision feedback equalizer (DFE) structure. Conver
ting this detector to a Bayesian scheme with an appreciation of non-li
near distortion in the read head modestly improves the velocity stabil
ity. The acquisition time for the channel speed is linear and rapid: a
50% change in velocity can be accommodated in fewer than 150 magnetic
Bur transitions. Results with real signals from a high-density magnet
ic credit card are provided.