Hear the voices
All three ship with the repo, and all three are spoken entirely on your own machine — no cloud, no API key, nothing leaving the computer.
Abby
Cute, slightly nerdy, young and warm. Calm, never in a hurry — for careful work, tricky debugging, and being talked through something gently.
Neya
Soft-spoken, emotionally intuitive, young and calming. Grounded and gently reassuring — for thoughtful deep dives, empathetic listening, and navigating complex ideas at a comfortable pace.
Max
Brave and driving, a trainer's energy. Punchy lines, counts off what is done, pushes for one more — for grinding through a long list.
English and
Russian are tested, and both sound
right — Russian included, though the voices were cloned from English. 🌐 Most widely
used languages should read normally too: the model underneath is multilingual, and
nothing in between cares which alphabet it is handed. A less widely spoken one may be
taken for its bigger neighbour, though — 
Bulgarian comes out fluent but in a Russian accent. Write a line in one
and listen.
Easy install
Windows only. Paste this into Claude Code:
Clone https://github.com/TraxData313/claude-voice here, then follow docs/tour.md in it: set it up, turn it on, and give me the spoken tour.
That is the whole of it. Claude installs it, opens the panel and switches it on — then tells you out loud how to drive it. A few minutes, nearly all of it download: a few hundred MB for the default engine, which runs on any PC.
Or in PowerShell, with no clone:
powershell -c "irm https://raw.githubusercontent.com/TraxData313/claude-voice/main/install.ps1 | iex"
Either way it becomes a program of its own: Abby for Claude in the Start menu, an Uninstall in Settings → Apps, a private copy of python.org's signed Python, no administrator rights anywhere, and fine with Smart App Control. What it does, and where it puts things →
Three engines
Pocket, the default, needs no graphics card at all and runs on any PC. Qwen is the original: it reads every alphabet and wants an NVIDIA card. Breeze acts as well as reads — it laughs, sighs and whispers where the text asks it to — and wants a 16 GB card. Switch between them from the panel whenever you like. What each costs and can do →
Games can speak through it too. Immersive AI for Mount & Blade II: Bannerlord gives every character in Calradia a voice of their own through it, installed right from the game. What a game needs →
What it wants: Windows, and a few hundred MB of disk for Pocket, the default engine, which runs on the processor. No administrator rights at any point, and nothing to sign up for. Qwen, which reads every alphabet, wants an NVIDIA card with roughly 4 GB of video memory, 3 GB of disk, and Smart App Control off — what it needs.
These pages exist because GitHub will not play audio inside a README — it offers the file for download instead. Everything else about the project lives in the repository.
