Researchers have introduced GigaEdge Spench, an open-source SDK designed for streaming speech enhancement across various platforms including servers, mobile devices, and web browsers. The SDK allows applications to construct enhancement pipelines using neural denoising, normalization, sample-rate conversion, and echo cancellation modules. It supports both 8 kHz INT8 denoisers and 48 kHz FP32 models, with a C++ core enabling integration with ONNX Runtime and TensorFlow Lite, as well as native support for Android, iOS, and WebAssembly. AI
IMPACT This SDK could streamline the deployment of advanced speech enhancement models across a wide range of devices and applications.
RANK_REASON The cluster describes a new software development kit (SDK) for speech enhancement, detailed in an arXiv paper. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →