FUNDAMENTAL FREQUENCY ESTIMATION FOR USE WITH A SINGING PITCHING DEVELOPMENT SYSTEM FOR PRIMARY SCHOOL CHILDREN
JAS ANGUS, PE GARNER, David M. Howard · 2024
The SINGAD (SINGing Assessment and Development) system enables the singing pitching skills of primary school children to be monitored and developed.Thissystem originally made use of a BBC microcomputer and the fundamental 'equency (f0) analysis was carried out by means of an analogue peak-picking circuit.Currently, an Atari microcomputer is used which depends on a commercially available pitch-to MIDI converter for its ] input.As with all tb estimation devices, these devices make their own idiosyncratic errors, but both generally perform well with children.There is a need to replace this pitch-to-MIDI converter for two reasons: rstly it is no longer available, and secondly there is a need to convert the system to operate on a PC computer where a MIDI input is imppropriate.We have developed an analogue f0 estimation device based on the earlier peak-picking circuit which outperforms the previously employed systems.This paper describes the basic operation of the device and provides a brief comparison between its output and the electrolaryngograph which is o en employed as a reference in such comparisons.Proceedings oi the lnstltute of Acoustics Fundamental frequency estimation for a singing pitching development system but this version of SINGAD has provided the data for a large study of the singing pitching skills of primary school children [7,8].Work is currently in progress to implement a new version of SINGAD for PC compatible computers.The voice lb input could be provided to a SoundBlaster card via a commercial pitchto-MIDI device.However, these are no longer available and in any case their use precludes a full understanding of their potential errors or any modi cation of the algorithm employed.Previous experience with the peak-picker suggested that this approach had a number of advantages for this application.There were however, situations when its output was not always robust, and this paper describes modi cation that have been made in an attempt to improve its overall operation for this application. BACKGROUNDResearch into the measurement of speech fa began in the 1920 s and still continues today.The