End-to-end realtime stack for connecting humans and AI

A Benchmark for Evaluating Turn-Taking and Overlap Handling in Full-Duplex Spoken Dialogue Models

τ-Bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains

High-Quality Voice Cloning TTS for 600+ Languages

A TTS that fits in your CPU (and pocket)

Lightning-fast, on-device TTS — running natively via ONNX.

Build local voice agents with open-source models

Code for MIRA: Multiplayer Interactive World Models with Representation Autoencoders

Open-source voice agent OS. Rust runtime, AI-driven builder, sub second latency. Self-host everything.

AI that sees your screen, listens to your conversations and tells you what to do

A Gemini 2.5 Flash Level MLLM for Vision, Speech, and Full-Duplex Multimodal Live Streaming on Your Phone

An awesome browser extension that reads aloud webpage content with one click

The open-source voice synthesis studio powered by Qwen3-TTS.

🎵 The Ultimate Open Source Suno Alternative - Professional UI for ACE-Step 1.5 AI Music Generation. Free, local, unlimited. Stop paying for Suno!

The most powerful local music generation model that outperforms most commercial alternatives

Qwen3-ASR is an open-source series of ASR models developed by the Qwen team at Alibaba Cloud, supporting stable multilingual speech/music/song recognition, language detection and timestamp prediction.

Qwen3-TTS is an open-source series of TTS models developed by the Qwen team at Alibaba Cloud, supporting stable, expressive, and streaming speech generation, free-form voice design, and vivid voice…

Your personal voice interface into any app. Speak naturally and your words appear wherever your cursor is, with fully customizable AI voice dictation. Open source alternative to Wispr Flow.

An open-source AI system enabling real time Quran recitation tracking, word-level alignment, error detection, and adaptive feedback. Designed to support memorization, accuracy, and accessibility fo…

The python library for real-time communication

An Open Source implementation of Notebook LM with more flexibility and features

A dataset of 100M connections between 5 different modalities.

On-device TTS model by Neuphonic

A package for NeuCodec: a 50hz, 0.8kbps, 24kHz audio codec.

A simple, performant and scalable JAX-based world modeling codebase

minimal reimplementation of deepmind's genie 3 world model

Qwen3-omni is a natively end-to-end, omni-modal LLM developed by the Qwen team at Alibaba Cloud, capable of understanding text, audio, images, and video, as well as generating speech in real time.