All people share implicit mappings across the senses, which give us preferences for certain sensory combinations over others (eg light colours are preferentially paired with higher-pitch sounds; Ward et al, 2006 Cortex 42 264-280). Although previous work has tended to focus on the cross-modality of vision with other senses, here we present evidence of systematic cross-modality between taste and sound. We created four sound continua representing to varying extents phonetic qualities of speech (F1, F2, voice discontinuity, and spectral balance). Sixty-five participants selected their preferred sound to accompany each of the basic tastes of sweet, sour, bitter, and salty, at two different concentrations. We found significant shared preferences among our participants to map certain acoustic qualities to certain types of tastes (eg sweet tastes tend to be mapped to a lower spectral balance than sour tastes). We also found a preference for mapping certain sound qualities to different taste concentrations. Together our data provide the first detailed analysis of how phonetic features map systematically to different tastants and concentrations. We examine the roots of these mappings, and discuss how such associations might guide the ways in which human languages are used to name objects with taste.