Projects sharing features with Moonshine

WhisperLiveKit is a real-time speech-to-text server that transcribes streaming audio into text with ultra-low latency using Whisper models. It serves transcription capabilities through REST endpoints and WebSocket connections, enabling external applications to send audio and receive transcriptions as words are spoken, making it suitable for live captioning or voice interfaces. The project distinguishes itself by combining real-time transcription with speaker diarization, assigning transcribed words to individual speakers during live audio streams for meeting or interview transcripts. It also

whisper.cpp is a C++ implementation of the Whisper speech-to-text model, serving as a lightweight machine learning inference engine and quantized runtime. It provides high-performance automatic speech recognition and real-time audio transcription without requiring a Python environment. The project utilizes model quantization to reduce memory usage and increase inference speed on local hardware. It incorporates hardware acceleration to optimize processing speed across different processors. The system covers audio processing capabilities including voice activity detection, speaker diarization,

espeak-ng is a multilingual text-to-speech engine and C-based library that converts written text into spoken audio across various languages, accents, and regional dialects. It functions as both a programmatic interface for embedding synthesis capabilities into external applications and a phonetic text converter that translates written text into phoneme codes. The system utilizes multiple synthesis methods, including formant synthesis to generate vocal sounds mathematically and diphone synthesis to produce audio by concatenating pre-recorded phonetic segments. It incorporates a speech processo

Auto-subs is an AI transcription and automatic captioning tool that converts spoken audio from video files into synchronized subtitles. It functions as a subtitle generator and a transcription bridge, enabling the conversion of speech to text with automatic speaker identification and multi-language translation support. The software prioritizes data privacy by utilizing on-device AI inference to process audio and video files locally on the user's hardware. It distinguishes itself by offering deep integration with professional video editing workflows, allowing users to export timing and transcr

ExecuTorch is a lightweight C++ runtime for deploying PyTorch models on mobile, embedded, and edge hardware. It provides an ahead-of-time compilation pipeline that exports, quantizes, and lowers model graphs into compact serialized programs, then executes them through a minimal runtime with hardware acceleration and on-device large language model inference capabilities. The project distinguishes itself through a hardware accelerator delegate system that partitions model subgraphs and offloads computation to specialized backends including NPUs, GPUs, and DSPs from Apple, Arm, Intel, MediaTek,

Paddle-Lite is a deep learning inference engine and edge computing runtime designed to execute trained models on mobile and edge devices. It provides a hardware-accelerated inference framework and a decoupled runtime with a minimal binary footprint to operate in resource-constrained environments without third-party dependencies. The project includes a model quantization tool for reducing precision and size via static and dynamic quantization, as well as a computation graph optimizer. These tools reduce latency and memory usage by fusing operators and pruning the model intermediate representat

PocketSphinx is an offline speech recognition engine that converts raw audio from files or live microphone streams into written text without requiring a network connection. It functions as a speech-to-text library, a real-time transcription engine, and a voice command processor, capable of detecting and transcribing spoken commands from continuous audio streams with configurable acoustic and language models. The engine uses weighted finite-state transducers to represent acoustic, phonetic, and language models as a single search graph for efficient decoding. It employs fixed-point acoustic mod

Neutts is a neural text-to-speech engine designed for real-time streaming output on edge devices such as phones and laptops. It supports voice cloning from short audio references, enabling zero-shot reproduction of a target speaker's voice, and can be fine-tuned or retrained from scratch for custom voices and styles. The system distinguishes itself through a decoder-only architecture that halves memory and accelerates generation on constrained hardware, combined with quantized model inference for reduced memory footprint. Its streaming decoder loop interleaves synthesis with playback, deliver

Whisper.cpp is a high-performance, local-first speech recognition engine designed to run large-scale machine learning models on consumer hardware. It functions as a portable library that converts audio into text, supporting both static file transcription and real-time stream processing. By utilizing a lightweight inference engine and weight quantization, the project minimizes memory and compute overhead, allowing for efficient execution without reliance on external cloud APIs or internet connectivity. The project distinguishes itself through a hardware-agnostic compute abstraction that offloa

Kokoro is a lightweight neural text-to-speech engine that converts written text into spoken audio using a compact model designed for fast inference. It supports multiple languages through language-specific grapheme-to-phoneme conversion pipelines, and offers voice profile selection to change the character of the generated speech. The engine provides GPU acceleration on Apple Silicon hardware by setting a single environment variable, enabling faster inference on Mac M-series machines. It also includes pattern-based text segmentation, allowing input text to be split at user-defined delimiters t

gemma.cpp is a C++ inference engine for Gemma, PaliGemma, and Griffin language models, designed to run directly on-device without Python dependencies. It provides a self-contained runtime that loads quantized model weights and performs text generation on CPU or GPU, along with a model checkpoint converter that transforms PyTorch or Keras checkpoints into a compact binary format for fast loading. The engine supports multiple model architectures, including the Griffin recurrent architecture with gated linear recurrent layers and sliding-window attention for efficient long-sequence handling, as

The nexa-sdk is an on-device AI SDK and multimodal inference engine designed to run large language, vision, and audio models locally on mobile and desktop hardware. It functions as a local LLM runtime and NPU acceleration framework, enabling the execution of generative and discriminative models without reliance on cloud services. The project distinguishes itself through a dedicated NPU acceleration framework that optimizes model execution on Neural Processing Units to reduce latency and power consumption. It employs hardware-agnostic backend routing to dynamically distribute computations acro

fastllm is a set of specialized software components for model weight conversion, Mixture-of-Experts runtimes, and tensor parallelism. It provides an OpenAI compatible API server to expose large language model capabilities through a standardized request format. The project features a tensor parallelism framework that splits computational workloads across multiple GPUs to accelerate execution. It includes a dedicated runtime optimized for Mixture-of-Experts architectures and a quantization tool to convert model weights into lower precision formats to reduce memory usage and increase throughput.

Kokoro-FastAPI is a text-to-speech API and LLM speech synthesis server that generates spoken audio from text via a REST interface. It functions as a Kubernetes-native deployment designed for orchestrated speech synthesis. The system includes a voice blending engine that creates unique vocal profiles by mixing multiple existing voices using custom weight ratios. The service provides real-time audio streaming to reduce latency and generates word-level timestamps for speech synchronization. It manages hardware efficiency through on-demand model loading to optimize VRAM usage and includes system

CTranslate2 is a C++ inference engine and runtime for Transformer models, designed to execute models on both CPU and GPU with optimizations for speed and memory efficiency. It functions as a model format converter, quantization tool, and REST API server, enabling deployment of neural machine translation, automatic speech recognition, and text generation models. The engine distinguishes itself through a suite of runtime optimizations including layer fusion, weight-matrix quantization, batch-by-length grouping, and a caching allocator that reuses GPU memory. It supports tensor-parallel model di

llama.cpp is a high-performance C++ inference engine and runtime for executing large language models locally across various hardware architectures. It provides the core components for local model execution, including a dedicated model quantizer for compressing weights into the GGUF format and a system for generating text embeddings for semantic search. The project distinguishes itself through specialized memory and execution optimizations, such as block-wise weight quantization to reduce memory footprints and memory-mapped model loading. It supports structured text generation by using formal

This project is an AI-powered IDE extension and LLM coding assistant that provides a conversational interface for generating, refactoring, and debugging code. It functions as an AI agent framework and a Model Context Protocol client, connecting AI models to external data sources and tools to automate complex development tasks. The system is distinguished by its use of autonomous AI agents capable of multi-step task execution, including the ability to read files, modify code, and run terminal commands iteratively. It supports recursive agent orchestration through subagent delegation and employ

This project is a neural text-to-speech system and voice trainer that converts written text into spoken audio across a variety of global languages and regional dialects. It functions as an ONNX-based engine capable of performing fast offline inference and uses a phoneme-based controller to manage precise pronunciation. The system distinguishes itself through a comprehensive toolkit for neural voice training, allowing for the creation of custom single-speaker or multi-speaker models. It supports the export of these models to a standardized open format and provides hardware acceleration via gra

Mace is a mobile deep learning inference framework and hardware acceleration engine. It functions as a runtime for executing neural network models on mobile devices, distributing computations across CPUs, GPUs, and NPUs. The project includes a cross-platform model converter for transforming pre-trained neural networks from various industry formats into mobile-optimized representations. It also provides a neural network obfuscator that converts model weights into source code to protect intellectual property from reverse engineering. The framework manages on-device resources by optimizing memo

AI-on-the-edge-device is an edge AI meter digitizer and computer vision image processor designed to convert images of analog and digital utility meters into numeric values. It functions as an IoT gateway that runs neural network inference locally on hardware to monitor water, power, and gas readings. The system is distinguished by its ability to handle both analog pointers and digital digits through custom-trained neural networks. It includes specialized tools for image alignment, region-of-interest extraction, and hardware-level lighting control to minimize glare on glass surfaces. To mainta

Whisper-diarization is a system for identifying and separating different speakers in audio recordings by combining OpenAI Whisper for transcription with automated speaker attribution. It functions as a pipeline that isolates vocal tracks from background noise and assigns transcribed segments to specific individuals. The project uses forced alignment to synchronize transcribed text timestamps with audio signals, improving the accuracy of speaker attribution. It employs voice activity detection to separate speech from silence and noise, ensuring precise boundaries for identification. The syste

Pyannote.audio is a PyTorch toolkit for speaker diarization, speaker identification, and speech activity detection. Its primary purpose is to partition audio recordings into segments and assign each segment to a specific speaker identity to determine who spoke when. The project includes a framework for classifying speaker identities and a pipeline for distinguishing human speech from background noise. It provides specialized tools for handling symmetric-overlap speech, where multiple speakers talk simultaneously, and employs learnable band-pass filters for raw waveform feature extraction. Th

Sherpa-ncnn is an edge-based speech recognition and synthesis engine designed to run neural network models locally on mobile, embedded, and desktop hardware. It provides a cross-platform framework for offline speech-to-text transcription and text-to-speech synthesis, ensuring that all audio processing occurs on-device without requiring an internet connection or external cloud services. The project distinguishes itself through its use of the ncnn inference engine, which is optimized for low-latency execution on resource-constrained devices. It incorporates on-device model quantization to reduc

This repository is a collection of sample applications and reference projects for integrating on-device machine learning APIs into mobile applications. It provides implementation examples for both Android and iOS, demonstrating how to embed computer vision and natural language processing models. The project covers the implementation of mobile computer vision, including text recognition, face detection, and barcode scanning. It includes workflows for real-time image processing and the integration of pre-trained models that process data locally on the device.

LiteRT-LM is a high-performance inference framework designed to execute large language models locally on mobile, desktop, and IoT hardware. It serves as an on-device model runtime that utilizes CPU, GPU, and NPU acceleration to provide low-latency processing. The framework is distinguished by its ability to process text, vision, and audio inputs through a single multi-modal inference engine. It features a local HTTP server that emulates OpenAI-compatible API endpoints and a WebGPU-based runtime for executing models directly within a web browser. To ensure output reliability, it includes a con

AidLearning-Framework is an integrated development platform for building and deploying AI applications on ARM-based devices. It combines Android and Linux operating systems running simultaneously on a single device, providing a unified runtime environment for cross-system AI development. The platform includes hardware acceleration across CPU, GPU, and NPU, with a unified API that automatically selects the optimal compute backend for inference. The framework distinguishes itself by enabling Python-based AI projects to be packaged directly into Android APK files for installation on mobile devic

Jetson Containers is a container management system that builds and runs GPU-accelerated Docker images for machine learning workloads on ARM64 edge hardware. It functions as a CUDA container orchestrator, automatically detecting the host's CUDA toolkit version and GPU capabilities to ensure container compatibility at runtime, while selecting the correct container image by matching the host's JetPack or L4T version at launch time. The project delivers pre-configured containers for executing quantized large language models and retrieval-augmented generation pipelines optimized for edge devices,