AMToBI

AMToBI, short for Autosegmental-Metrical ToBI, is a framework for annotating the prosody of spoken language by combining autosegmental-metrical (AM) phonology with the ToBI labeling approach. It provides a language-general architecture for representing how pitch accents, phrase accents, and boundary tones are aligned with syntactic and prosodic structure. By modeling prosody as a hierarchy of tonal events linked to prosodic boundaries, AMToBI supports cross-language annotation while allowing language-specific realizations.

In AMToBI, annotators typically label a set of tonal events and boundary markers on dedicated annotation tiers. Common tonal labels include pitch accents (for example, high or low targets) and their contours, while boundary tones mark the end of prosodic units. Break indices or boundary indicators denote the edges of prosodic phrases. The AM framework enables these events to be associated with abstract prosodic units such as intonational phrases and intermediate phrases, capturing multi-layered patterns of rise and fall in intonation.

Applications of AMToBI span linguistic research, speech technology, and corpus annotation. It has been applied to multiple languages, with language-specific AMToBI schemes developed to reflect local prosodic phenomena. The framework supports analyses of intonational contrasts, prosodic structuring of utterances, and their perceptual/functional roles. It is also used in developing and evaluating speech synthesis and recognition systems, where accurate prosody modeling is important. Resources typically include annotation guidelines, corpora with AMToBI labels, and tools to assist with annotation and analysis.

See also: ToBI, autosegmental-metrical phonology.