Awesome GitHub RepositoriesOn-Device Speech Recognizers

This project is a Python speech recognition library that serves as a unified interface for converting spoken audio into text. It functions as a bridge between Python applications and a variety of speech-to-text engines, providing a consistent way to interact with both local and cloud-based recognition services. The library distinguishes itself as a multi-engine transcription tool, wrapping diverse online APIs and offline recognition backends into a standardized format. This allows for interchangeable recognition engines and supports multilingual audio transcription through various language pa

Cactus 是一个端侧 AI 推理引擎，专为在移动和可穿戴硬件上执行大语言模型、视觉模型和语音转文字系统而设计。它提供了一个用于定义矩阵运算和激活函数序列的可编程张量计算图，以及一个利用本地文本文件为模型响应提供依据的本地检索增强生成（RAG）框架。 该项目具有一个多平台 SDK，包含用于将 AI 功能集成到移动应用程序中的语言绑定，以及一个将外部模型格式转换为优化本地执行的模型转换系统。它利用混合路由系统，根据硬件容量在端侧执行和云端提供商之间重定向工作负载。 该引擎涵盖了广泛的功能面，包括用于语音活动检测和转录的端侧音频处理、用于相似性搜索的向量嵌入生成，以及用于将模型输出解析为外部函数调用的工具集成。这些过程由针对移动硬件低延迟性能优化的原生内核提供支持。

Porcupine is an on-device wake word detection engine that listens for a specific spoken phrase in real-time audio and triggers actions, all processed locally without any cloud connectivity. It includes a custom wake word model creator that generates production-ready models from just a few spoken examples in seconds, requiring no training data. Beyond wake word detection, Porcupine also provides on-device speech recognition for real-time transcription with custom vocabulary, an on-device audio content searcher that indexes and finds spoken phrases in audio files or streams, and a lightweight vo

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

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

Tesseract-OCR-iOS 是 Tesseract 引擎在 iOS 应用程序中的原生集成。它提供设备端图像识别功能，用于识别并从图像中提取打印文本，将其转换为可编辑字符串。 该项目支持本地设备执行，这意味着图像处理和文本提取完全在硬件上进行，无需使用外部云服务或网络请求。它利用 C++ 包装器和 Objective-C 桥接器，将高级 iOS API 与底层的 Tesseract 引擎和 Leptonica 图像处理库连接起来。 该库支持文档数字化工作流，以及开发从相机或照片读取文本的辅助功能工具。它直接从应用程序包中管理特定语言训练数据文件的加载。

Tess-two is an optical character recognition tool and Android application designed to extract written text from images using the Tesseract engine. It functions as an image analysis utility for detecting visual artifacts, blur, and optical flow within local image files on Android devices. The project includes an image pre-processing suite used to clean and manipulate images to increase the accuracy of text recognition. This involves a pipeline that applies grayscale conversion and binarization before the recognition process. The software integrates native image processing and character analys

Foundry-Local 是一款机器学习开发工具，旨在促进私有的、设备端推理和模型管理。它提供了一个本地服务器环境，直接在用户的硬件上托管机器学习模型，确保所有数据处理（包括提示处理和音频转录）都保留在本地环境中，而无需外部云连接。 该项目通过自动化整个模型生命周期（包括资产的发现、下载和版本控制以保持与主机硬件的兼容性）而脱颖而出。它具有一个硬件抽象层，可自动检测并选择最高效的可用处理器进行计算密集型任务，从而无需手动配置即可实现硬件加速执行。 除了核心推理外，该工具还包括一个用于交互式模型探索和性能验证的命令行界面。它还提供标准化的 API 代理，使用行业标准协议将传入请求映射到本地模型端点，以支持与外部软件框架的集成。

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

该项目是一个全面的工具包，用于设备端语音识别、合成和音频处理，专为 Apple Silicon 工程设计。它提供了一个框架，用于构建完全离线运行的实时、全双工语音代理，利用原生硬件加速来保持性能和隐私。通过利用优化的机器学习模型，该库实现了复杂音频任务的本地执行，而无需依赖外部云服务。 该库通过其对本地、高性能语音交互的专门关注脱颖而出。它包括用于流式音频流水线的复杂编排，允许以低延迟进行实时转录、语音合成和语音克隆。该系统旨在处理持续的、交互式的对话，具有内置机制来防止音频反馈循环并管理持久的流会话。 除了核心交互外，该项目还提供了一套广泛的音频增强和管理功能。它支持高级信号处理，包括源分离、降噪和音频上采样，以及用于说话人日志记录和嵌入提取的工具。该框架还提供广泛的模型管理工具，例如量化控制、内存管理和对自定义模型权重加载的支持，确保开发者能够在本地硬件上平衡处理速度和资源消耗。 该项目包含一个用于执行音频任务和将模型权重转换为优化格式的命令行接口。它还暴露了 HTTP 和 WebSocket 端点，以促进与标准行业接口的集成。

Expo Speech Recognition is a cross-platform mobile module that converts live microphone audio and pre-recorded files into text using native speech engines. It provides offline speech recognition capabilities by downloading and verifying local speech models to enable on-device processing without an active network connection. The library includes session lifecycle management to start, stop, or abort recording, alongside real-time spoken language detection with confidence scoring. It emits volume change events for metering interfaces, handles audio session configuration and routing, and persist