Enhancing the Stability of LLM-based Speech Generation Systems through Self-Supervised Representations

Open in new window