ToBI

ToBI, short for Tone and Break Indices, is a framework for annotating prosody in spoken language. It was developed in the 1990s to provide a standardized method for labeling intonation and phrase boundaries, facilitating comparison across speakers and languages. The original English ToBI system has since spawned a family of language-specific variants used in linguistics and speech technology.

The core idea of ToBI is to separate tonal information from structural boundaries. The tonal tier encodes pitch events such as pitch accents and boundary tones. Common pitch accents in English ToBI include markers like H* and L* to indicate high and low prominences, while boundary tones such as H% or L% signal the end of an intonational phrase. A separate break-index tier encodes boundary strength between phrases on a scale typically ranging from 0 (no boundary) to 4 (a strong boundary). Additional tiers may capture phrase structure or other contextual information.

ToBI is widely used for linguistic analysis, cross-dialectal comparison of prosody, and applications in speech technology, including speech synthesis and automatic speech recognition. It supports systematic annotation of prosodic patterns, aiding research into focus, question types, discourse structure, and language processing.

There are multiple language adaptations within the ToBI framework, such as EN ToBI for English and J_ToBI for Japanese, among others. While the approach provides a standardized method, annotating ToBI transcripts requires training and careful listening, as there can be variation in labeling across dialects and speakers.