Two unit selection singing synthesisers

摘要

Two speech synthesisers were adapted for singing synthesis using unit selectionudtechniques provided by the Festival speech synthesis system. A limited domainudapproach was used by focussing on the pitch, duration and word of each note.udThe first synthesiser used the cluster unit technique on a database of an octaveudrange, where each note had a specific word assigned to it. Some of the automaticudtechniques used (e.g. for segmentation) were designed for speech and shouldudideally be adapted to take account of the differences between singing and speaking.udBetter quality was achieved with a multisyn engine and improved database design.udThis database used a smaller pitch range and only three syllables, ’la’ ’ti’udand ’so’, but each syllable could be synthesised on any available note, and in anyudcombination of notes and syllables. This was achieved by weighting the targetudcost of selecting units from the database in favour of choosing units with the correctudpitch and duration. Finally, prosodic modification was applied to units inudthe multisyn engine, but this degraded quality as a result of how the units wereudmodified.udAlthough the quality of synthesis was appropriate for the intended applications,udthe database was small and linguistic structure simple. To build a larger scaleudsinging synthesiser, either some aspect of the database should be kept simple,udsuch as vocabulary, or prosodic modification of units should be improved throughudfurther analysis of the characteristics of singing.