音声
MeloTTS
MeloTTS — multilingual text-to-speech for English, Spanish, French, Chinese, Japanese and Korean.
★ 7.7k スターMIT
ガイドを見る →maryttsは、maryttsが公開しているオープンソースの音声プロジェクトです（GitHub: marytts/marytts）。GitHubスターは2,582、フォークは771で、NOASSERTIONライセンスの下で公開されています。GitHubでの最終更新は2025年1月17日です。
MaryTTS — open-source, multilingual text-to-speech platform written in pure Java
Decide whether you will run the standalone server (easiest from the releases page) or build from source / embed MaryTTS in a Java project.
Download a ready-to-use package from the GitHub releases page, or clone the source repo.
From source, run it with Gradle; optionally start the voice installer GUI to download more voices, then restart the server.
Open the server at http://localhost:59125, or send it an HTTP synthesis request from Python/shell using the linked txt2wav examples.
The whole client-server system is written in pure Java and runs wherever a JVM runs.
Designed as a multilingual TTS platform with downloadable voices; the default artifact referenced in the README is a US English HSMM voice.
Any programming language can synthesize by compiling, starting the server and querying it over HTTP (examples for Python 3 and shell are linked).
./gradlew runInstallerGui downloads and installs additional voices (restart the server to use them).
Declare a dependency on MaryTTS voice artifacts to use it inside your own Java projects.
Gradle tasks (distZip/distTar/installDist) build a standalone server distribution, and user dictionaries can extend pronunciations.
Clone MaryTTS (or download a ready package from the releases page instead).
git clone https://github.com/marytts/marytts.git
cd maryttsCompile and test all modules and create outputs under build/.
./gradlew buildStart a MaryTTS server and open it in your browser (port and address are configurable with -Dsocket.port/-Dsocket.addr).
./gradlew run # http://localhost:59125The marytts open-source project maintains it; it originated from DFKI's MARY platform (older docs at mary.dfki.de and the marytts-wiki).
Yes — the README states the code is under the LGPL version 3 license. (GitHub's license field shows NOASSERTION; the README/license file states LGPL-3.)
No — synthesis runs locally on your own server; there is no cloud API involved.
pyttsx3 wraps your OS's voices in Python and gTTS calls Google Translate's endpoint. MaryTTS is a standalone multilingual server platform you host, with its own downloadable voices and an HTTP API for any language.
Download the release package or clone and run ./gradlew run, then open http://localhost:59125.
MeloTTS — multilingual text-to-speech for English, Spanish, French, Chinese, Japanese and Korean.
Private cross-platform speech-to-text with LLM cleanup built in.
Chatterbox — Resemble AI's open-source TTS family: expressive text-to-speech and voice cloning, multilingual V3, low-latency Turbo and tiny Nano models, MIT
Native Linux dictation: fast, private, and hackable speech-to-text.
Free AAC communication boards with text-to-speech, in the browser.
Voice cloning inside Kokoro: natural multilingual speech, your voice.
LibreAIカタログに掲載された音声プロジェクトの中で最もスターの多いもの — maryttsが合わなかった場合に検討したい、現実的な候補リストです。
利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.
Robust Speech Recognition via Large-Scale Weak Supervision
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
Clone a voice from 5 seconds of audio and generate arbitrary speech in real time
Port of OpenAI's Whisper model in C/C++
カタログの中でmaryttsとスター数が近い音声プロジェクト — 前後の棚もそのまま見て回れます。
gTTS — Python library and CLI that turns text into spoken mp3 via Google Translate's speech API
Offline text-to-speech for Python — uses your operating system's own voices, no internet needed
Vietnamese TTS with instant voice cloning — on-device, real-time CPU inference, 48 kHz audio
VoiceMem — streaming dual-brain memory for voice agents: a left brain for facts, a right brain for emotion
MARS5 speech model from CAMB.AI — clone a voice from 5 seconds of audio with striking prosody
GAN vocoder that turns mel-spectrograms into high-fidelity speech audio, fast enough for real-time use.