JAM is a compact rectified-flow-based lyrics-to-song generation system designed for fine-grained word- and phoneme-level vocal timing control. It provides pretrained-model inference with audio or text style conditioning, configurable song duration, and a complete training pipeline covering pretraining, supervised fine-tuning, and Direct Preference Optimization (DPO).

Movie Narrator is an open-source Python toolkit for automatically producing narrated movie recap videos from a single command. It uses LLMs to research and generate scripts, text-to-speech for narration, subtitle generation and translation, scene matching, background music, configurable rendering templates, video QA, and local or remote asynchronous task processing through a CLI and REST API.

OpenGuider is an open-source Electron desktop AI assistant for completing real user-interface tasks on a local machine. It combines multi-provider LLM chat, structured planning and orchestration, screenshot-aware guidance with pointer coordinates, optional speech input/output, session persistence, global shortcuts, and an extensible plugin system featuring browser automation through browser-use.

CC-Beeper is a native macOS menu bar and floating-widget application that monitors Claude Code sessions and alerts users when tasks finish, fail, require permission, or need input. It provides animated status states, global hotkeys, configurable auto-accept modes, local voice dictation and read-aloud features, themes, sounds, session tracking, and local HTTP-based Claude Code hooks.

wsay is a Windows command-line text-to-speech application equivalent to macOS's `say`. It uses Windows Speech API to synthesize speech, select voices and playback devices, read input from text files, clipboard, pipes, or interactive mode, save WAV files, support Speech XML and Unicode, and apply configurable volume, speed, pitch, and radio-style audio effects. It also provides a reusable C++ library API for synchronous and asynchronous speech playback.

Textreader Pro is a web-based text-to-speech emulator that converts user-entered text into audio using Amazon Polly voices and Google Wavenet. It provides a Gatsby/React frontend with voice selection, audio playback, light/dark/system themes, usage logging, rate limiting, and browser-based proof-of-work bot protection, backed by an Express API.

audio.cpp is a high-performance, cross-platform native C++ framework for local audio and multimodal inference, built on ggml. It provides shared CLI, server, WebUI, model-management, batching, streaming, pipeline, and testing infrastructure for numerous TTS, ASR, voice-conversion, diarization, VAD, source-separation, music-generation, alignment, and audio-processing models, with GGUF support and optimized GPU execution.

OpenReels is an open-source AI video production pipeline that transforms a topic into a publish-ready vertical MP4 for YouTube Shorts, TikTok, or Instagram. It automates research, scriptwriting, voiceover, AI and stock visuals, music, animated captions, transitions, rendering, cost estimation, and AI quality critique, with access through a CLI, REST API, live-progress web UI, and Docker deployment.

AivisSpeech Engine is a Japanese text-to-speech engine based on VOICEVOX ENGINE, integrated into the AivisSpeech desktop application. It provides a VOICEVOX-compatible HTTP API, supports AIVMX ONNX voice models using Style-Bert-VITS2 architectures, and enables expressive speech synthesis through parameters such as emotion strength and tempo dynamics. It runs on CPU or supported GPU backends and is distributed for Windows, macOS, Linux, and Docker.

GhostCut, now branded internationally as Weydub, is an AI-powered video and image localization platform. It provides automated subtitle and text removal with OCR and inpainting, hard-subtitle translation, speech translation and dubbing using ASR/TTS, video remixing, music matching, and image translation, with API integration support. The repository mainly contains product documentation, demos, and a Python example for calling the video translation API.