Generation of speech with a formant synthesizer such as the Klatt KLSYN88 (Klatt and Klatt, 1990) requires that a large number of parameters (KL parameters) be manipulated. Many of these parameters are interrelated, and considerable precision may be required in specifying them. This paper describes a method of control of a formant synthesizer using higher-level (HL) parameters that are more independent of each other. These parameters are directly related to articulatory and aerodynamic variables, and have the advantage that they can be specified with less precision than the KL parameters.