Synaesthesia is an unusual phenomenon, in which additional sensory perceptions are triggered by apparently unrelated sensory or conceptual stimuli. The main foci of this thesis lie in speech sound - colour and voice-induced synaesthesia. While grapheme-colour synaesthesia has been intensively researched, few studies have approached types of synaesthesia based on vocal inducers with detailed acoustic-phonetic and colorimetric analyses. This approach is taken here. First, a thorough examination of speech-sound - colour synaesthesia was conducted. An experiment is reported that tested to what extent vowel acoustics influence colour associations for synaesthetes and non-synaesthetes. Systematic association patterns between vowel formants and colour measures could be found in general, but most strongly in synaesthetes. Synaesthetes also showed a more consistent pattern of vowel-colour associations. The issue of whether or not speech-sound - colour synaesthesia is a discrete type of synaesthesia independent of grapheme-colour synaesthesia is discussed, and how these might influence each other. Then, two experiments are introduced to explore voice-induced synaesthesia. First, a comprehensive voice description task was conducted with voice synaesthetes, phoneticians and controls to investigate their verbal voice quality descriptions and the colour and texture associations that they have with voices. Qualitative analyses provided data about the nature of associations by the participant groups, while quantitative analyses revealed that for all groups, acoustic parameters such as pitch, pitch range, vowel formants and other spectral properties influenced colour and texture associations in a systematic way. Above all, a strong connection was found between these measures and luminance. Finally, voice-induced synaesthetes, other synaesthetes and controls participated in a voice line-up, of the kind used in forensic phonetic case work. This experiment, motivated by previous findings of memory advantages in synaesthetes in certain areas, tested whether synaesthetes’ voice memory is influenced by their condition. While no difference in performance was found between groups when using normal speech, voice-induced synaesthetes outperformed others in identifying a whispering speaker. These are the first group studies on the otherwise under-researched type of voice-induced synaesthesia, with a focus on acoustic rather than semantic analysis. This adds knowledge to the growing field of synaesthesia research from a largely neglected phonetic angle. The debate around (re)defining synaesthesia is picked up. The voice description experiment, in particular, leads to a discussion of a synaesthesia spectrum in the population, as many common mechanisms and associations were found. It was also revealed that less common types of synaesthesia are often difficult to define in a rigid way using traditional criteria. Finally, the interplay of different types of synaesthesia is discussed and findings are evaluated against the background of the existing theories of synaesthesia.