116 repos across 5 sub-areas
Libraries, models, and deployment tools for automatic speech recognition (ASR), with heavy emphasis on running inference efficiently across platforms. The cluster centers on OpenAI's Whisper model and its derivatives—whisperX, openasr, whisper-diarization—alongside production-focused implementations in Python, Rust, and Swift for edge devices and real-time processing. Repositories span model optimization (ONNX, quantization), speaker diarization, mobile/iOS deployment, and open-source ASR alternatives to commercial services.
Automatic Speech Recognition (ASR)
32 repos
Python-based libraries, models, and frameworks for converting spoken audio into text using deep learning approaches. The cluster covers end-to-end ASR systems, model implementations (particularly with PyTorch and ONNX), training pipelines, and deployment-ready solutions. Includes both general-purpose transcription tools and specialized variants like multilingual and robust ASR models designed for real-world audio conditions.
Whisper-based speech recognition
31 repos
Python-focused tools and applications built around OpenAI's Whisper speech-to-text model, including implementations for real-time transcription, speaker diarization, video processing, and optimized inference (like whisper.cpp for C++ deployment). The cluster spans from research notebooks and command-line utilities to production inference frameworks, covering both the core Whisper ecosystem and downstream applications that depend on it.
Whisper Speech Recognition & Deployment
25 repos
Speech-to-text systems built around OpenAI's Whisper model, with a focus on efficient inference, cross-platform deployment, and real-time transcription. The cluster spans multiple implementation languages—Python for data processing and service layers, Rust and Swift for performance-critical and native platform implementations—reflecting the breadth of use cases from backend services to desktop and mobile applications. Repositories range from Whisper bindings and C++ optimizations (whisper-cpp) to end-user tools like real-time transcription HUDs and conversation recording utilities.
Automatic Speech Recognition & Audio Processing
18 repos
Speech-to-text and audio processing systems built primarily with Python and deep learning frameworks. The cluster centers on ASR model implementations and optimizations, including fine-tuned variants of Whisper and other speech recognition architectures. You'll find model compression techniques (distillation, quantization), language-specific adaptations, and inference optimizations across inference frameworks like JAX, alongside foundational audio processing and feature extraction tools.
Speech Recognition and On-Device Audio Processing
10 repos
Libraries and SDKs for speech-to-text and speech recognition, with emphasis on neural network-based models and on-device processing. The cluster centers around Deepgram's multi-language SDK implementations (Node.js, JavaScript, Go, Python, .NET) alongside complementary tools for audio processing, machine learning inference, and speech model deployment. Developers here work with both cloud-based and edge-deployed speech recognition systems, with particular attention to real-time transcription and latency-sensitive audio applications.