ProsodyLM: Uncovering the Emerging Prosody Processing Capabilities in Speech Language Models
Paper
β’
2507.20091
β’
Published
This repository contains the model checkpoints and sample training data for
the paper ProsodyLM: Uncovering the Emerging Prosody Processing Capabilities in Speech Language Models.
llm/: ProsodyLM checkpoint and tokenizertts/: TTS checkpoint and speaker embeddingsdata/: A small-scale sample dataset (same format as the real training data)If you use this resource, please cite the paper above.
License: CC BY-NC 4.0