VocalNet: Speech LLM with Multi-Token Prediction for Faster and High-Quality Generation
This addresses the need for efficient, real-time voice interaction systems, representing an incremental improvement through a novel method applied to a known bottleneck.
The authors tackled the problem of slow and low-quality speech generation in speech large language models by introducing VocalNet, which uses multi-token prediction to achieve faster and higher-quality generation, with experiments showing it performs on par with mainstream models and surpasses existing open-source ones.
Speech large language models (LLMs) have emerged as a prominent research focus in speech processing. We introduce VocalNet-1B and VocalNet-8B, a series of high-performance, low-latency speech LLMs enabled by a scalable and model-agnostic training framework designed for real-time voice interaction. Central to our contribution is the first application of multi-token prediction (MTP) to speech LLMs. This approach represents a paradigm shift from standard next-token prediction (NTP), offering simultaneous improvements in generation speed and quality. Informed by analysis of MTP's effect on speech generation and experimental comparisons, we designed a straightforward and highly effective MTP implementation. Experiments demonstrate that VocalNet performs on par with mainstream Omni LLMs even with limited training data, and significantly surpasses existing open-source speech LLMs. To foster reproducibility and community advancement, all model weights, inference code, training data, and framework implementations have been made publicly available at https://github.com/SJTU-OmniAgent/VocalNet