# moonshine | Moonshine AI | Knowledge Base | Instagit

Fast and accurate automatic speech recognition (ASR) for edge devices

GitHub Stars: 3.8k

Repository: https://github.com/moonshine-ai/moonshine

---

## Articles

### [How Moonshine's C++ Core, C API, and Language Bindings (Python, Swift, Java) Are Structured](/moonshine-ai/moonshine/moonshine-cpp-core-language-bindings-structure)

Explore Moonshine's layered architecture: C++ inference engine, stable C API, and Python Swift Java bindings. Discover how language bindings interact with the C++ core for efficient inference.

- Tags: internals
- Published: 2026-02-16

### [What Is Ergodic Streaming in Moonshine’s Real-Time Transcription?](/moonshine-ai/moonshine/moonshine-ergodic-streaming-explained)

Discover Moonshine's ergodic streaming for real-time transcription. Achieve sub-200ms latency by processing audio frames once and reusing cached data for efficient transcription.

- Tags: deep-dive
- Published: 2026-02-16

### [Moonshine C API Handle-Based Object Management for Cross-Language Compatibility](/moonshine-ai/moonshine/moonshine-c-api-handle-based-object-management)

Learn how Moonshine's C API uses handle-based object management for cross-language compatibility. Discover thread-safe maps and lifetime management for seamless integration.

- Tags: internals
- Published: 2026-02-16

### [Moonshine Streaming State Management for Incremental Decoding: Architecture and Implementation](/moonshine-ai/moonshine/moonshine-streaming-state-management-components)

Explore Moonshine's streaming state management for incremental decoding. Discover how Transcriber and Stream classes manage state and dispatch real-time transcript events.

- Tags: architecture
- Published: 2026-02-16

### [How Moonshine's Streaming Architecture Achieves Low-Latency Speech Recognition vs. Whisper](/moonshine-ai/moonshine/moonshine-streaming-latency-vs-whisper)

Discover how Moonshine's streaming architecture achieves sub-500ms speech recognition latency by processing audio in small chunks outperforming Whisper's batch processing.

- Tags: architecture
- Published: 2026-02-16