AivisSpeech

📖 What is AivisSpeech?

An AI voice synthesis engine that excels at emotionally rich speech. It runs fast on CPU alone, and its VOICEVOX-compatible interface makes switching over easy. You can add voice models with one click from the official model-sharing platform “AivisHub,” and adjust speaker style, speed, tempo, and volume per sentence to add emotional expression. It also supports a custom dictionary for names and proper nouns, and a VOICEVOX-compatible engine API (with Yukkuri MovieMaker 4 integration). The core app is LGPL-3.0 open source, free with no credit required (usage rights follow each voice model's own license).

💡 When to Use It

A great fit when you want expressive synthesized voices for chat readouts or commentary banter, for free. Since the UI is VOICEVOX-compatible, if you're already using VOICEVOX you can add more character voices without relearning anything. For video editing, it integrates with Yukkuri MovieMaker 4 to build narration for explainer videos. Because it runs on CPU alone, it's easy to use even on setups where the GPU is busy encoding the stream. Just be sure to check each voice model's own license before using it.

📋 Basic Info

🚀 How to Use

1. Step 1: Download the installer (or Zip) version of AivisSpeech from aivis-project.com or GitHub, and set it up.
2. Step 2: On first launch, the model data downloads automatically — wait for it to finish.
3. Step 3: Pick the voice model you want on AivisHub, and install it with one click while AivisSpeech is running.
4. Step 4: Type in your text, adjust speaker style, speed, and tempo per sentence, check the read-aloud result, and export it as a WAV.
5. Step 5: To use it on stream, call the VOICEVOX-compatible engine API from a tool like OneComme, or play the exported audio through an OBS media source.

🎬 Tutorial Video

🎉 Highlights & Comments