AUDIOCOMMAND SPEECH
Instruction and Command Voice Sessions
Spoken instructions and short follow-up exchanges that preserve intent, correction, confirmation and clarification behaviour.
View datasetAUDIOPROSODY
Expressive and Styled Speech Sets
Speech changes with pace, emphasis, delivery and context even when the words remain the same. Expressive and Styled Speech Sets make those requested delivery conditions visible so acoustic variation can be connected to the prompt, speaker and style instruction that produced it.
This is more useful than treating expressive speech as one broad category. Comparable prompts can be captured across distinct speaking styles, allowing technical teams to study how models respond to controlled differences in delivery without implying that a style label represents a person's internal emotional state.
Technical structure
- Prompt identity linked to transcript and speaker metadata
- Project-defined style or delivery instructions
- Paired or comparable utterances where useful to the design
- Optional prosodic descriptors such as pace, emphasis or timing-related fields
Typical applications
Prosody modelling, expressive speech synthesis, style-conditioned models and speech robustness research.
Delivery note
Style labels describe the requested delivery condition and are defined for the project.
Delivery
The exact package depends on the collection scope. Where relevant, delivery can include task-specific records, manifests, provenance fields, stable identifiers, SHA-256 hashes, MLCommons Croissant 1.0 metadata and loading instructions.