This paper presents a detailed description of the multimedia corpus that was built for the phonetic analysis of Russian ironic speech. The corpus was developed after a series of preliminary studies. We analysed the expression of irony in films, series, audiobooks and other material presented in the open sources. Special attention was given to the contexts, in which the antiphrasis (irony as negation) appeared. Analysis of their lexical content, semantic structure and phonetic characteristics allowed us to construct the material for reading, which was the closest to the natural circumstances of irony expression. 60 Russian native speakers read the sets of short dialogues and coherent texts. Audio recording was simultaneously accomplished with video recording. The same target fragments were inserted in various types of sentences in ironic and non-ironic contexts. These homonymous fragments were extracted from the recorded material. Orthographic, prosodic and phonetic annotation was done. It included information about the context, target fragment transcription, stressed syllable and stressed vowel boundaries, intonation model, co-occurring paralinguistic phenomena and perceptual evaluation of irony by native listeners. A case study of phonetic properties of ironic speech showed that ironic utterances are characterized by the contrast in intensity level and stressed vowel duration as compared with non-ironic utterances, the different usage of intonation models in the two types of utterances was also observed.