Speak — let your AI talk out loud

Gives your OpenCode AI a voice. Instead of only reading its answers, you hear them.

Easiest: automatic install (Windows, 1 step)

Copy this line into PowerShell, press Enter, and follow what it says:

powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/humbleangel/opencode-tool-speak/main/install-windows.ps1 | iex"

(Keep reading below only if you prefer to install by hand.)

What you need (all free)

1. Python — download it from python.org. On Windows, tick the box "Add python.exe to PATH" during installation.

2. A sound player — the easiest is ffmpeg (it includes ffplay): download it from ffmpeg.org and add it to PATH. VLC also works.

3. One small library — open a terminal and run:

   pip install edge-tts

Setup (about 2 minutes)

1. Copy these 3 files into your OpenCode tools folder:
   • speak.py, speak.ts, speak.json
   • Windows: C:\Users\YOUR-NAME\.config\opencode\tools\
   • Mac/Linux: ~/.config/opencode/tools/
2. Restart OpenCode.
3. Done. On its very first message, the agent will tell you out loud that it can speak and listen, and ask if you want that always on.

How to use

• Just chat — the agent speaks on its own.
• To change the voice, ask in plain words, e.g. "use a British female voice".
• Voices are grouped by language (en-US-... English, fr-FR-... French, and so on). If the accent sounds wrong, the voice doesn't match the language of the text — ask for a voice in your language.

If something goes wrong

• No sound? Install ffmpeg, then close and reopen your terminal (so it picks up the new PATH) and restart OpenCode.
• Error mentioning edge-tts? Run pip install edge-tts again.
• Strange pronunciation of words with accents? The voice language must match your text — ask for a voice in your language. (Technical note: this tool forces the speech request to use the voice's own language, which stock setups get wrong.)

License

MIT — free for everyone, see LICENSE.