Can music be rated consistently using nonverbal descriptors such as colours and temperatures? 144 participants rated 6 experimenter-selected and 2 self-selected pieces of music along 15 bipolar icon (graphic) scales intended to portray emotions, and sensory experiences consisting of colour, temperature, shape, speed, texture, and weight. Participants also rated the same pieces using bipolar verbal scales which aimed to encompass the concepts represented by the icons (e.g., the word “red” for the colour red). Furthermore, the icons themselves were subjected to open-ended verbal labelling to validate the icon scale. Colour icons spontaneously evoked a cross-modal association on 67% of occasions: blue being cool, and red/orange being warm or hot, and the icon scale had overall good face validity. Music regularly and consistently evoked multisensory associations (using the icon scale) including shapes, colours, weight, and temperatures, in addition to emotions. Cross-modal perception is indicative of music’s character rather than the enjoyment of the music. The icon scale provides new insights into music perception and for applications where language skill may limit participant expression.