real-time
10 repositories
Self-hosted team chat built for security-first organizations — Rocket.Chat is an open-source, TypeScript-based communications platform offering self-hosted, cloud, or air-gapped deployment, end-to-end encryption, and role-based access control. It's used by organizations like Deutsche Bahn and the US Navy that need full control over sensitive communications, making it a fit for teams wary of proprietary chat tools like Slack.
Open Source framework for voice agents, multimodal apps, and realtime AI. Maintained by Daily and the community.
Incremental indexing engine keeps agent context fresh — CocoIndex is a Rust-core, Python-exposed engine that re-syncs only the changed delta from codebases, Slack, meeting notes and docs into vector stores or knowledge graphs. It's built for teams running long-horizon AI agents or RAG pipelines that need continuously up-to-date context instead of full reindexing on every change.
A streaming query engine for data grids, pivots, and charts in any runtime — Perspective pairs a drag-and-drop grid, pivot table, and WebGL charting UI with a C++ query engine compiled to WebAssembly, Python, and Rust. Notable extras include a 64-bit memory64 build for in-browser datasets over 4GB, OPFS/mmap paging to disk, DuckDB support, and Jupyter and React integrations for embedded analytics.
SeaTunnel extends data integration to multimodal and LLM workloads — Apache SeaTunnel syncs huge data volumes across 160+ connectors, now covering video, images, and unstructured text alongside structured data for embeddings and LLM pipelines. It runs on Zeta, Flink, or Spark engines, supporting batch, streaming, and CDC with a distributed snapshot algorithm for consistency. Useful for teams needing flexible, multi-engine data pipelines.
Open-source Vietnamese TTS with instant voice cloning, now 48kHz — VieNeu-TTS v3 Turbo is a from-scratch Vietnamese TTS stack with its own codec and phonemizer, now 48kHz, 25 preset voices, experimental emotion tags, and batched multi-speaker conversation mode. It runs on-device with real-time CPU/GPU inference, while the more accurate voice-cloning v4 is kept proprietary via API, making this repo the open option for developers and researchers.
Compose video streams in real time with React components — Smelter lets developers mix live video and audio from multiple sources, adding text, shaders, and embedded websites using React components. Built in Rust by Software Mansion, it targets low-latency, programmable streaming and broadcasting workflows. The repo also bundles gpu-video, an open-source hardware video coding library.
CPU-only multilingual live transcription that beats Whisper on Japanese CER — hayamimi routes each utterance to a specialist quantized ONNX model per language via sherpa-onnx instead of one general Whisper model, hitting 3.8% CER on Japanese broadcast audio vs Whisper-large-v3-turbo's 13.8%, at 10-50x realtime on a 6-core CPU, under 2GB RAM, with no GPU or cloud API.
Real-time generative rendering turns any reference video into game-ready visuals — EdenLive decouples interactive world state (run by engines or coding agents) from appearance, which it renders in real time guided by a first frame and text prompt. It hits 24+ FPS at 480P/720P on a single consumer GPU, with open 5B/14B weights, streaming inference code, and an EdenGolem skill that turns a prompt into a playable game prototype.
Your console. Your GPU. Smoother gameplay. Open-source real-time frame generation & upscaling for console gaming.