valentinIA
Zero-token native neural voice assistant for AI coding sessions in VS Code & Antigravity IDE.
valentinIA is a zero-token native IDE voice assistant that recites AI agent text responses word-for-word out loud in natural, studio-quality 24kHz neural voices. It runs 100% locally on your machine with zero LLM prompt token consumption, zero cloud API fees, and zero visual UI clutter.
⚡ Quick Action: Launch General Settings
Click the button above or execute in Command Palette (Cmd+Shift+P / Ctrl+Shift+P):
👉 valentinIA: General Settings
Key Capabilities
- Zero-Token Local Execution: Operates natively in Node.js/TypeScript inside the IDE process. No LLM prompt context, tokens, or cloud API costs.
- Standalone C++ Neural Engine: Spawns lightweight, local C++ Piper neural synthesis processes without Python, pip, or virtualenv requirements.
- Host & Workspace Isolation: Zero cross-over audio. Antigravity IDE and VS Code sessions remain 100% isolated.
- 1-Click Instant Mute: Clicking the status bar item (
valentinIA: Active) instantly mutes audio and terminates playback processes.
- Audited Female Voice Catalog: Studio-quality female voice models for 30+ regional accents and languages.
- Natural Speech Cadence: Automatic Markdown punctuation naturalization for headers, lists, and horizontal section dividers.
Voice Catalog & Regional Accents
This extension includes audited studio-quality voice variants for:
- English: USA (
LJ Speech High), UK (Cori High)
- Spanish: Argentina (
Daniela Studio 24kHz), Spain (MLS Female), Mexico (Claude High)
- Portuguese: Brazil (
Faber Medium)
- French: France (
Siwis Medium)
- German: Germany (
Kerstin Low)
- Italian: Italy (
Paola Medium)
- Other Languages: Europe (Dutch, Polish, Ukrainian, Swedish, Danish, Finnish, Greek, Czech, Hungarian, Romanian, Turkish, Catalan), Asia (Chinese, Japanese, Korean, Hindi, Vietnamese, Thai), Middle East (Arabic, Hebrew, Persian), Africa (Swahili, Amharic, Yoruba, Hausa).
Quick Setup & Usage
1. General Settings Webview
Open the interactive onboarding panel anytime via:
- Command Palette (
Cmd+Shift+P / Ctrl+Shift+P) ➔ valentinIA: General Settings
- Status Bar Menu ➔ Click
valentinIA: Active ➔ General Settings
2. Status Bar Control
$(unmute) valentinIA: Active: Click to INSTANTLY MUTE speech output for the active session.$(mute) valentinIA: Muted: Click to ACTIVATE voice output.
Extension Settings
valentinIA contributes the following configurable settings:
| Setting |
Default |
Description |
valentinia.enabled |
true |
Enable or disable native voice notifications globally. |
valentinia.voice |
"en_US-ljspeech-high" |
Active regional female voice model key. |
valentinia.speed |
0.85 |
Speech synthesis duration rate (0.75x, 0.85x Default, 0.95x, 1.05x). |
Privacy & Local Security
valentinIA operates 100% offline and locally on your machine. Speech synthesis takes place inside ~/.valentinIA/ using native C++ executables. No audio data or conversation text is ever transmitted over the network.
Author & License
Created and maintained by marcedelgado.dev (Marcelo Delgado).
Distributed under the MIT License. Contributions are welcome on GitHub.
| |
