Abstract: A voice service node (VSN) serving a wireless speech terminal (WST) connected to a mobile switching center (MSC) comprises a bypass unit, a speech codec and a speech recognizer. The bypass unit exchanges messages with a remote bypass unit at the MSC, and switches the operation of the VSN between a PCM mode, for analog speech terminals, and a codec bypass mode, for digital speech terminals. The speech codec operates during the bypass mode to decode a speech frame into a sequence of digital words for the speech recognizer and to encode a voice prompt signal form a speech prompting system from a sequence of digital words into a speech frame. The speech recognizer receives the sequence of digital words and a flag indicating the status of the frame and provides a set of parameters characterizing the sequence of digital words. The set of parameters and the flag are applied to a distance computation unit for determining a set of distances between a speech model and the set of parameters.