EElevenLabs

VoiceStudio is an open-source desktop application for generating and working with synthetic speech locally. Its documented workflows include text-to-speech, zero-shot voice cloning, voice design, dictation, transcription, video dubbing, audiobook creation from EPUB or PDF files, and multi-voice stories. It supports multiple TTS and ASR engines, batch queues, speaker diarization, vocal isolation, watermarking, and GPU-aware engine routing.

The application is designed for user-controlled deployment on macOS, Windows, and Linux, with downloadable DMG, MSI, and AppImage releases plus Docker installation guidance. Processing runs on the user’s machine without accounts, API keys, or cloud dependence, while a remote-backend option is also documented. The project is licensed AGPL-3.0 and exposes an OpenAI-compatible API and MCP server.

EmotiVoice is an open-source text-to-speech engine that generates English and Chinese speech with more than 2,000 voices. Its defining workflow uses prompts to control emotion and style, including happy, excited, sad, and angry delivery. The README also documents personal-data voice cloning, adjustable voice speed, an interactive web interface, batch scripting, and an OpenAI-compatible TTS API.

Users can run the provided Docker image with an NVIDIA GPU, install the project through Conda and pip, or use the Streamlit demo page locally. The project includes pretrained-model setup and command-line inference workflows, and an HTTP API is available for application integration. It is released under the Apache-2.0 license.

Abogen is an open-source text-to-speech application for converting EPUB, PDF, text, Markdown, and subtitle files into audio with synchronized captions. It supports audiobook creation, voiceovers, adjustable speech speed, voice previews, custom voice mixing, chapter selection, separate or merged chapter files, and batch queues.

The project provides a PyQt6 desktop GUI and a Flask web UI, with downloadable WAV, FLAC, MP3, OPUS, and M4B outputs plus SRT or ASS subtitles. It uses Kokoro-based models, supports Windows, macOS, and Linux installation, and documents pre-downloading models and voices for offline use. The web interface runs locally at port 8808.

GPT-SoVITS is an open-source voice synthesis application with a WebUI for zero-shot and few-shot text-to-speech, voice conversion, and voice cloning. It can use a five-second vocal sample for instant synthesis or about one minute of training data for fine-tuning, with support for Chinese, English, Japanese, Korean, Cantonese, and related workflows.

The project includes tools for audio separation, dataset segmentation, multilingual ASR, transcription proofreading, model fine-tuning, inference, and speech-speed control. It can be installed on Windows, Linux, and macOS, run through Python and Conda, or deployed with Docker Compose; pretrained models and optional ASR and UVR5 models are downloaded separately.

ChatTTS is an open-source generative speech model designed for dialogue scenarios such as LLM assistants. It synthesizes expressive speech in English and Chinese, supports multiple speakers, and provides control over prosodic elements including laughter, pauses, and interjections. Streaming audio generation and sampled speaker embeddings are documented capabilities.

The repository includes a WebUI, command-line inference, Python usage examples, and audio export through torchaudio. It can be installed from PyPI, GitHub, or a local checkout and run with local Python dependencies and suitable hardware; the README cites at least 4GB of GPU memory for a 30-second clip. The code is AGPLv3+, while the released model is CC BY-NC 4.0 for academic and research use.

ebook2audiobook converts ebooks and short text into audiobooks using multiple text-to-speech engines, including XTTSv2, Bark, Fairseq, VITS, Tacotron2, Tortoise, GlowTTS, and YourTTS. It supports many ebook and document formats, OCR for image-based pages, chapter and metadata handling, multilingual synthesis, optional voice cloning, custom or fine-tuned models, SML controls for pauses and voice switching, and output formats such as M4B, MP3, WAV, FLAC, and OGG.

The project can run locally on Linux, macOS, and Windows through a Gradio web interface or headless commands. It documents Docker, Hugging Face Spaces, Google Colab, and Kaggle execution paths, with CPU, CUDA, ROCm, MPS, XPU, and Jetson options. It is intended for use with legally acquired, non-DRM ebooks.

Voice-Pro is an open-source AI multimedia web application for creators, researchers, and multilingual professionals. Its workflows combine YouTube downloading and audio extraction, vocal separation, speech recognition, translation, subtitle generation, voice cloning, and text-to-speech in a Dubbing Studio. It supports Whisper variants, F5-TTS, E2-TTS, CosyVoice, kokoro, and Edge-TTS, with multilingual processing and configurable speech output.

The Gradio WebUI includes Dubbing Studio, Whisper Caption, Translate, and Speech Generation tabs. Users can process common media formats, export WAV, FLAC, or MP3 audio, create subtitles, translate subtitle files, and perform real-time voice recognition and translation. The repository describes Windows support, especially with an NVIDIA GPU, and provides a portable installer using start.bat; Mac and Linux operation is unverified.

Chatterbox TTS Server runs Resemble AI’s Chatterbox model family behind a modern Web UI and OpenAI-compatible API. Users can enter text, choose among Original, Multilingual, and Turbo engines, adjust generation parameters, use predefined voices or reference audio for cloning, and generate expressive speech with tags such as [laugh] and [cough]. It supports 23 languages through Chatterbox Multilingual and can process long text by chunking and joining segments for audiobook generation.

The project is self-hostable and documents automated local installation, Windows portable mode, Google Colab use, and Docker deployment. It runs with NVIDIA CUDA, AMD ROCm, Apple Silicon MPS, or CPU fallback, and exposes API endpoints including streaming text-to-speech and voice listing. The documented scope centers on speech synthesis and voice cloning; transcription, dubbing, music, sound-effects, and the

Alexandria Audiobook converts books and novels from TXT, Markdown, or EPUB into fully voiced audiobooks. An LLM annotates scripts with speakers, dialogue, and delivery directions, while the built-in Qwen3-TTS engine generates multilingual speech. Users can assign preset, cloned, designed, or LoRA-trained voices, preserve speaker aliases, adjust individual lines, regenerate selected chunks, preview sequences, and review scripts with a second LLM pass.

The browser-based editor provides a five-step workflow for setup, scripting, voice assignment, rendering, and results. Exports include combined MP3, per-line MP3, Audacity multi-track files, and chaptered M4B. It runs locally through Pinokio, Google Colab, or Docker with NVIDIA GPU support, and can connect to local or cloud OpenAI-compatible LLM services.

viXTTS Demo is a Vietnamese voice-cloning and text-to-speech tool based on a fine-tuned XTTS-v2.0.3 model and the viVoice dataset. Users can enter text, generate speech in Vietnamese, and use cloned voices; the README also says other languages are offered, though their effectiveness has not been tested.

The project provides a hosted Hugging Face Space and a local Ubuntu or WSL2 workflow. Local use installs dependencies through run.sh, launches a Gradio demo, and saves generated results in an output directory. It recommends at least 16GB of RAM, 10GB of disk space, and an Nvidia GPU with 4GB of VRAM, while CPU inference is supported but slower.