---
title: README
emoji: 🎧
colorFrom: indigo
colorTo: gray
sdk: static
pinned: false
---

# Fullduplex

Fullduplex tracks full-duplex and speech-to-speech voice AI: the models that listen and speak at the same time, the datasets they are trained on, and the benchmarks that test them.

- Website: [fullduplex.ai](https://www.fullduplex.ai)
- [Models](https://www.fullduplex.ai/models), [Datasets](https://www.fullduplex.ai/datasets) and [Benchmarks](https://www.fullduplex.ai/benchmarks), one page per entry
- [Analytics](https://www.fullduplex.ai/analytics): hosted APIs and published Full-Duplex-Bench scores side by side
- [What is full-duplex voice AI?](https://www.fullduplex.ai/knowledge) ([日本語](https://www.fullduplex.ai/ja/knowledge))
- [Weekly report](https://www.fullduplex.ai/signals) on new papers, models and datasets

The whole catalogue is on the Hub as [Fullduplex-ai/catalog](https://huggingface.co/datasets/Fullduplex-ai/catalog).