Search papers, labs, and topics across Lattice.
The paper introduces DuplexCascade, a VAD-free cascaded ASR-LLM-TTS pipeline that enables full-duplex speech-to-speech dialogue by converting long turns into chunk-wise micro-turn interactions. This approach leverages conversational special control tokens to manage turn-taking and response timing within streaming constraints, allowing for rapid bidirectional exchange. Experiments on Full-DuplexBench and VoiceBench demonstrate that DuplexCascade achieves state-of-the-art full-duplex turn-taking and maintains strong conversational intelligence compared to other open-source systems.
Unlock full-duplex speech-to-speech dialogue without VAD limitations using chunk-wise micro-turns and special control tokens to steer LLM behavior in a cascaded pipeline.
Spoken dialog systems with cascaded ASR-LLM-TTS modules retain strong LLM intelligence, but VAD segmentation often forces half-duplex turns and brittle control. On the other hand, VAD-free end-to-end model support full-duplex interaction but is hard to maintain conversational intelligence. In this paper, we present DuplexCascade, a VAD-free cascaded streaming pipeline for full-duplex speech-to-speech dialogue. Our key idea is to convert conventional utterance-wise long turns into chunk-wise micro-turn interactions, enabling rapid bidirectional exchange while preserving the strengths of a capable text LLM. To reliably coordinate turn-taking and response timing, we introduce a set of conversational special control tokens that steer the LLM's behavior under streaming constraints. On Full-DuplexBench and VoiceBench, DuplexCascade delivers state-of-the-art full-duplex turn-taking and strong conversational intelligence among open-source speech-to-speech dialogue systems.
