The universal dictation input apparatus and method disclosed employs multiple wireless dictation devices (“WDD”) and base stations each of which include a 900 MHz radio for communicating with one another over selected radio frequency (“RF”) channel pairs to conduct dictation sessions on centralized dictation systems (“CDSs”). Each base station is uniquely assigned a single channel pair. The WDDs are programmed with code representing the exclusive channels of a select number of base stations. When a user takes a WDD off-hook, a WDD sequentially and cyclically scans for carrier transmitted by authorized base stations until an idle base station is identified and linked to over the authorized channel. The user of a WDD must be authorized to conduct dictation sessions on one or more CDSs. To this end the WDDs includes a dialing address keypad to address authorized CDSs. The CDSs are third party systems addressed from a WDD by DTMF dialing digits. Dictation sessions are conducted by sending dictation commands, for example “record” and “play” in the form of DTMF digits to the CDSs. The WDDs send data as ASCII codes in packets to a base station. The transmitted data includes dialing digits and system commands for the base station. The WDDs generate data from the dialing keypad, programmable keys customized to send frequently used dictation commands to a CDS and to provide other specialized features and a barcode reader. Specific setup data is downloaded to a base station to enable it to execute commands and dialing, digits unique to CDS of different vendors and to prepare for receipt of verbal dictation commands. A base station includes a voice recognition algorithm to translate verbal commands spoken at a WDD, such as “record” and “play,” into codes forwarded to a CDS as DTMF digits.