@NielsRogge: Just added this blog as a project page to https://paperswithcode.co/paper/2410.00037…! Hope that more people can learn …
Summary
NielsRogge added a blog explaining the Moshi full-duplex voice model as a project page on Papers With Code, aiming to increase accessibility to the state-of-the-art architecture.
View Cached Full Text
Cached at: 06/18/26, 06:10 PM
Just added this blog as a project page to https://t.co/6yQIMR6Ltn!
Hope that more people can learn about state-of-the-art full-duplex voice models this way :) https://t.co/IcZwVCjIp2
rohit (@bicro_): Moshi is one of the best open source full-duplex voice models out there. The architecture is dense, so we spent a few days studying it and wrote up what we learned, with diagrams to make it click faster.
Let us know if it was helpful 🤠
Similar Articles
@NielsRogge: Added web search to Papers with Code's upcoming chat interface This runs on @huggingface Inference Providers via @baset…
Niels Rogge adds web search to Papers with Code's upcoming chat interface, powered by Hugging Face Inference Providers, Baseten's DeepSeek-V4 Flash, Exa's MCP server, and Modal. He invites HF usernames for feedback.
@MosiAI_Official: MOSS-Transcribe-Diarize-0.9B is now open source on @huggingface. Built with an end-to-end audio-to-structured-transcrip…
MOSS-Transcribe-Diarize-0.9B is an open-source end-to-end audio understanding model for long-form multi-speaker transcription, diarization, and timestamp generation, released by Mosi AI under Apache 2.0.
@kyutai_labs: New paper: Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models We use RL to post-train speech models (Mo…
Kyutai Labs released a new paper on using reinforcement learning to post-train speech models (Moshi and PersonaPlex) for more human-like interaction, including when to respond, wait, or give listening cues.
@NielsRogge: Introducing a revival of PapersWithCode! As @ilyasut said, we're back to the "age of research". Hence, it's important t…
NielsRogge announces a revival of PapersWithCode, featuring SOTA per domain, leaderboards, and methods parsed at scale using AI agents.
MoshiRAG: Asynchronous Knowledge Retrieval for Full-Duplex Speech Language Models
MoshiRAG combines a compact full-duplex speech language model with asynchronous retrieval-augmented generation to improve factuality while maintaining real-time interactivity. The approach leverages natural temporal gaps in conversation to retrieve external knowledge without disrupting the natural flow of dialogue.