Speakrs Full PyAnnotate pipeline in Rust/ONNX 20-37x times faster macOS
AI/MLSpeakrs is a Rust-based implementation of the full PyAnnotate diarization pipeline (segmentation, clustering, VBx) that runs entirely without Python, using ONNX Runtime or native inference. It achieves 20–37x faster performance on macOS, making it ideal for developers and ML engineers who need high-speed, production-ready speaker diarization without Python overhead. Its interest lies in demonstrating that complex audio ML pipelines can be radically accelerated and deployed as lightweight, standalone binaries.
Cross-platform signals
You might also like
More in AI/ML
Self-hosted AI workspace.
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.