llms.txt
# Vois - The Voice AI Studio
> Professional voice AI that runs on your desktop. Create speech with 100+ voices, clone voices from approved samples, design new voices, and finish audio in one local production app. Subscriber includes unlimited classic production. Pro adds Omni, Voice Design, and 30 languages. Paid plans have no per-character fees or generation caps.
## About Vois
Vois is a fully-featured, local-first AI voice studio for desktop. Text-to-speech, voice cloning, voice design, editing, mastering, and export run on the user's machine. Setup, verified model-package downloads, software updates, activation, and periodic license validation require internet access. Connected features use the network when selected; desktop diagnostics are on by default; you can turn it off in Settings.
Website: https://vois.so
Homepage: an unlimited AI audiobook, podcast and voiceover generator that runs on the user's computer, with three guided workflows for audiobook creators, podcasters and voiceover creators. Generation is unlimited with no token limits because Vois runs on the user's computer; paid plans add unlimited exports. Journey states on the same page: https://vois.so/?start=audiobooks (narrator casting, chapter editing, export to Google Play Books, Kobo and Spotify), https://vois.so/?start=podcasts (host and co-host voices, episode editing, export presets for Spotify, Apple Podcasts and YouTube) and https://vois.so/?start=voiceovers (video and YouTube narration, tutorials and courses, game characters and explainers; export audio for your video editor, your course platform or your game engine). The independent audience pages are https://vois.so/for/audiobooks and https://vois.so/for/podcasters. The included library has 100+ multilingual voices with auditions in 23 languages. Generate locally, edit scripts, arrange tracks and export audio files. Export presets do not publish directly to platforms. Compatible local agents can operate Vois through the skills at https://vois.so/skills; ordinary web chats do not control the app directly.
---
## Key Features
### 100+ Natural Voices
- 100+ built-in natural voices; Pro unlocks Omni coverage across 30 languages
- Core Multilingual engine: 23 languages, including Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish
- 21 categories: Heroes, Villains, Companions, NPCs, Creatures, Machines, Narrators, Hosts, Announcers, Executives, Educators, Storytellers, Broadcasters, Mentors, Youth, Audiobook, Podcast, Explainer, App Assistant, Interactive, Expressive
- Speed control from 0.5x to 2.0x
- 24kHz high-quality audio output
- Learn more: https://vois.so/features/voices
### Voice Cloning
- Clone a permitted voice from a 10 to 15 second clip
- Create unlimited custom character voices
- Consent confirmation required for ethical use
- Paid feature: available on Subscriber and Pro (export credits unlock export only, not cloning)
- Voice cloning runs locally; reference audio is not uploaded for cloning
- Learn more: https://vois.so/features/voice-cloning
### Local Processing & Private Content
- Text-to-speech, editing, mastering, and export run on the user's machine
- Scripts, reference audio, and generated audio are not uploaded for TTS or mastering
- Setup, verified model-package downloads, software updates, activation, and periodic license validation require internet access
- Voice-engine packages download from cdn.vois.so and are checked against the release's verified manifest before use
- After an app update, Vois automatically re-downloads current packages only for voice engines previously installed; other engines remain on demand
- Model quality is user-selectable: Balanced (recommended) uses less memory and is faster; High quality uses more memory and is available on capable machines
- Local production can continue for up to 1 day after a successful license check while the license remains active
- Subscriber and Pro include unlimited generation with no per-character fees; the Free plan includes up to 10 generations per day
- Smart re-render: edits only regenerate the sentences that changed; unchanged lines return instantly at no extra cost
- Learn more: https://vois.so/features/offline
### Content Provenance
- Exported audio includes metadata identifying it as AI-generated by Vois
- Metadata contains a unique export ID, app version, and voice type (preset or cloned)
- Cloned voice generations are logged locally for audit trail
- Server-side export receipts provide forensic traceability for cloned voice exports
- No personal information, input text, or audio content is included in provenance metadata
- Uses locally bundled and downloadable voice models with clear licensing
### Optional Diagnostics
- Desktop diagnostics are on by default; you can turn it off in Settings
- When enabled, Vois sends redacted error reports, sanitized native crash stacks, and a coarse hardware profile; the same setting allows that profile to accompany license checks
- Reports can include error category, engine category, execution method, model quality profile, text length, optional numeric generation seed, timing, and coarse application state
- Diagnostics never include synthesis text, transcripts, voice names, absolute user paths, audio or PCM, license keys, or raw error messages; raw crash dumps are deleted locally and the local Vois log is never uploaded
- Unsent reports may remain queued in app storage for up to seven days; disabling diagnostics closes reporting, purges the on-disk and browser queues, and stops the native crash helper
- Sentry retains received reports for its standard event-retention period; the license service separately uses Sentry for redacted server-side errors
### Voice Library
- 100+ universal library voices across 21 categories
- Voice presets for any content type
- Organized by production use case, character role, accent, and listening style
- Learn more: https://vois.so/features/voices
### Multi-Speaker Production
- Assign different voices to different speakers
- Auto-detect speakers from script tags
- Color-coded speaker management
- Inline pause nodes for precise silence control (300ms to 3s): dramatic beats, scene transitions, chapter breaks
- Perfect for podcasts, audiobooks, and dialogues
- Learn more: https://vois.so/features/multi-speaker
### Professional Audio Export
- Platform presets include YouTube and podcast loudness targets; audiobook delivery must be matched to the distributor's current loudness, peak, noise-floor, file, and narration rules
- Built-in mastering pipeline with EQ, de-esser, and limiter
- Export to WAV, MP3, FLAC, AAC
- Learn more: https://vois.so/features/audio-export
### Pronunciation Dictionary
- Custom pronunciation rules for names, brands, acronyms, and technical terms
- CSV import/export for batch vocabulary management
- Global or project-specific scope
- Whole-word matching for accurate replacements
- Learn more: https://vois.so/features/pronunciation-dictionary
### 30-Language Multilingual Support
- Generate speech in 30 languages on Pro with Omni, using cloned voices, designed voices, and the library voices that support Omni
- Same voice identity across every language: clone in English, use in Japanese
- Complete list of all 30 Omni languages: https://vois.so/features/multilingual
- Mix languages within a single project using speaker tags
- Right-to-left script support for Arabic and Hebrew
- All processing runs locally, no per-language API fees
- Learn more: https://vois.so/features/multilingual
### CLI & AI Automation
- Local CLI command groups cover status, credits, generation, projects, voices, export, models, settings, and shell completions
- Control Vois from the command line or AI agents (Claude Code, Cursor, etc.)
- Batch generation, project management, export with mastering profiles
- JSON output mode for pipeline integration
- The desktop app, CLI process, and agent terminal must be on the same machine and running as the same operating-system user
- Agents must obtain user approval before state-changing commands; checkout and payment remain human-completed
- CLI generation accepts an optional numeric seed for reproducible Expressive or Multilingual takes
- Learn more: https://vois.so/features/cli-automation
### Website WebMCP Discovery
- `get_vois_agent_setup` returns public Agent Skill, installer, and local handoff information
- `get_vois_purchase_options` returns public subscription and credit-pack information with human checkout links
- Website WebMCP is read-only discovery. It does not install or launch Vois, create checkout, complete payment, or control the desktop app
- Real product control uses the installed Vois CLI on the same machine and operating-system user as the running app, through launch-token-authenticated local IPC
**Note on AI Audiobook Distribution:**
Distribution rules vary by platform and can change. Check each distributor's current AI-narration policy before production. ACX's Voice Replica beta is invitation-only and is not the same as accepting arbitrary externally generated narration.
### Listen Mode
- Convert articles, books, and documents to audio
- Supports PDF, EPUB, DOCX, Markdown, plain text
- Web article extraction from URLs
- Perfect for accessibility, commuting, or multitasking
### Background Music & Sound Effects
- Browse a curated audio library for timeline production
- Music and sound effects download on demand
- Downloaded assets are cached locally for reuse
---
## Who Is Vois For?
### For Creators
**Podcasters** - Solo hosts who want a full crew sound. Create multi-speaker shows, assign different voices to guests, export directly to Spotify and Apple Podcasts.
- Learn more: https://vois.so/for/podcasters
**Audiobook Authors** - Writers who want to narrate their own books. Turn manuscripts into finished audiobooks with professional narration quality. Export to Google Play Books, Kobo, and Findaway Voices.
- Learn more: https://vois.so/for/audiobooks
- Note: Audiobook acceptance rules vary by distributor and can change. Verify each platform's current AI-narration policy before production; ACX's Voice Replica beta is invitation-only.
**YouTube Creators** - Faceless channels, tutorials, explainers. Generate consistent voiceovers at scale without recording.
- Learn more: https://vois.so/for/youtube
**Documentary Makers** - Multi-language narration, historical character voices. Professional documentary-grade audio.
- Learn more: https://vois.so/for/documentaries
**Listen Mode Users** - Convert any text to audio for accessibility, commuting, or multitasking.
- Learn more: https://vois.so/for/listen
### For Industries
**Training & E-Learning** - Scale training without scaling teams. Create consistent professional training content across courses, languages, and updates. Perfect for L&D teams.
- Learn more: https://vois.so/for/training
**Advertising & Marketing** - A/B test voices, not just copy. Generate ad variations in seconds. Test different voices, tones, and scripts without booking talent.
- Learn more: https://vois.so/for/advertising
**Game Dialogue** - Voice every character in your game. Hundreds of NPCs, each with a distinct voice. No casting calls, no recording sessions.
- Learn more: https://vois.so/for/game-dialogue
**Accessibility** - Natural TTS for assistive technology. Screen reader quality voice generated locally; your text and audio stay on your device.
- Learn more: https://vois.so/for/accessibility
---
## Technical Specifications
### Platforms
- macOS 12+ (Apple Silicon)
- Windows 10/11 (64-bit)
### Requirements
- 16 GB RAM recommended
- 20 GB free disk space (voice engines download separately: about 5 GB for Multilingual, about 11 GB for all four)
- GPU optional (speeds up generation 2-6x on Apple Silicon)
---
## Pricing
Vois uses a **tiered subscription model**:
- **Free**: Free forever, for hobby, research and non-commercial use only; commercial use needs Subscriber or Pro. Email verification required to start (no credit card). Use Sonic, and try Fast, Expressive, Multilingual, Omni and Craft (generate and listen) in the app with 10 generations/day, 1 project, and 3 Listen items. Export up to 10 files a month of English audio made with Sonic (the Free allowance, reset on the 1st of each month, UTC). Purchased export credits are separate and export audio from any engine or language. No voice cloning, Voice Design, or general CLI access; `vois speak --follow --play` works on Free and plays speech aloud. Holding purchased credits also unlocks a pay-as-you-go CLI subset.
- **Subscriber**: Unlimited classic production, including generation and export
- **Pro**: Subscriber features plus Omni, Craft and Voice Design
- Current monthly and annual prices: https://vois.so/pricing
### Historical Checkout Note
Legacy Vois 2.x Product Hunt launch codes may still work at checkout: `PHVOIS2` for Subscriber and `PHVOIS2PRO` for Pro. Availability can change, so anyone looking for an older Product Hunt rate can try the relevant code and let Polar confirm whether it is currently accepted.
**Subscriber includes:**
- 100+ studio-quality voices
- Voice cloning & customization
- Core TTS engines (Sonic, Fast, Expressive, and legacy Multilingual)
- Unlimited audio generation and export
- Professional mastering & EQ
- Platform export presets
- Pronunciation dictionary
- CLI access for automation
- Script editor with multi-speaker support and inline pause nodes for precise silence control
**Pro adds:**
- Omni with 30 languages
- Craft: direct any voice with a recording (library, cloned with a transcript, or designed) in 10 languages; cloned voices keep the accent of their recording
- Voice Design: create new reusable voices from a description
- Direction in plain language: describe how a passage should sound
Cancel anytime. Your generated audio and projects remain on your device.
### Export Credits (Pay-as-You-Go)
Don't want a subscription? Free plan users can purchase export credits as one-time payments to download audio. See https://vois.so/pricing for current pack sizes and pricing.
- One credit = one audio export (download) with any voice
- Credits never expire and are separate from the 10 Free exports each month
- Subscribers don't need credits (unlimited exports included)
- Free plan: generate and preview audio for free, export up to 10 English Sonic files a month, and purchase credits to export other engines, languages or more files
- Pay-as-you-go = the Free plan plus export by credit. Voice cloning and Voice Design stay on the paid plans (Voice Design on Pro), and credits do not change the Free plan's non-commercial terms.
---
## Site Map
### Feature Pages
- /features - All features overview
- /features/voices - 100+ Natural Voices
- /features/voice-cloning - Voice Cloning
- /features/offline - Local Processing & Private Content
- /features/multi-speaker - Multi-Speaker Production
- /features/audio-export - Audio Export
- /features/pronunciation-dictionary - Pronunciation Dictionary
- /features/multilingual - 30-Language Multilingual Support
- /features/voice-design - Omni Voice Design
- /features/cli-automation - CLI & AI Automation
- /omni-gpu-support - Omni Hardware & GPU Support
### Use Case Pages - Creators
- /for/podcasters - For Podcasters
- /for/audiobooks - For Audiobook Authors
- /for/youtube - For YouTube Creators
- /for/documentaries - For Documentary Makers
- /for/listen - Listen Mode
### Use Case Pages - Industries
- /for/training - Training & E-Learning
- /for/advertising - Advertising & Marketing
- /for/game-dialogue - Game Dialogue
- /for/accessibility - Accessibility
### Free Resources & Newsletter
- /voice-direction-cheatsheet - Voice Direction Cheat Sheet: 27 script tricks that make AI narration sound human (free PDF)
- /audiobook-production-kit - Audiobook Production Kit: chapter batch plan, casting worksheet, pacing cheats, QC checklist (free PDF)
- /podcast-voice-kit - Podcast Voice Kit: season batch plan, intro formulas, voice selection worksheet (free PDF)
- /ai-voice-cost-calculator - AI Voice Cost Calculator: model your true cost per finished minute, plus a free pricing report
- /faceless-voice-system - Faceless Channel Voice System: 30-videos-a-month batch production planner (free PDF)
- /game-dialogue-kit - Game Dialogue Kit: pipeline checklist, line-batch spreadsheet, localization worksheet (free download)
- /newsletter - Studio Notes: the Vois newsletter, two short studio emails a week
### Blog
Latest articles on voice technology, creator tutorials, and product updates:
- /blog - Blog index with all articles
- /changelog - Latest Vois release notes
#### Creator Guides
- /blog/youtube-ai-voice-disclosure-rules - YouTube's AI Disclosure Rules for Voiceover Channels: What Actually Needs a Label
- /blog/ai-search-show-pages-creators - AI Search Now Reads Your Show Page: How to Write One That Gets Cited
- /blog/where-to-publish-ai-narrated-audiobooks - Where You Can Publish an AI Narrated Audiobook in 2026, Platform by Platform
- /blog/help-center-articles-customer-support-audio - Turn Help Center Articles Into Customer Support Audio
- /blog/video-game-voice-localization-workflow - A Practical Video Game Voice Localization Workflow
- /blog/private-audio-briefings-internal-communications - Create Private Audio Briefings for Internal Communications
- /blog/youtube-video-localization-10-languages - How to Localize a YouTube Video Into 10 Languages Without Reshooting
- /blog/batch-producing-podcast-season-weekend - Batch-Producing a Full Podcast Season in a Weekend
- /blog/acx-audiobook-compliance-ai-voices - ACX Audiobook Requirements and AI Narration: What to Know
- /blog/blog-post-to-youtube-narration-repurposing - From Blog Post to YouTube Narration: Repurposing Written Content With AI Voices
- /blog/ai-voiceover-saas-product-demos - AI Voiceover for SaaS Product Demos and Screencasts
- /blog/building-consistent-voice-brand-podcast - Building a Consistent Voice Brand for Your Podcast or YouTube Channel
- /blog/multi-track-timeline-ai-voice-editing - The Multi-Track Timeline: Editing AI Voice Productions Like a Pro
- /blog/faceless-youtube-no-api-costs - How to Build a Faceless YouTube Channel Without Per-Character API Costs
- /blog/ai-voiceover-corporate-training-no-cloud - AI Voiceover for Corporate Training Without Cloud Dependencies
- /blog/manuscript-to-audiobook-ai-production-workflow - From Manuscript to Audiobook: A Complete AI Production Workflow
- /blog/script-formatting-features-guide - The Complete Guide to Vois Script Formatting
- /blog/faceless-youtube-blueprint - The Faceless YouTube Blueprint: Build a Channel Without Showing Your Face
- /blog/automated-podcast-factory - The Automated Podcast Factory: Script to Published Episode in Under an Hour
- /blog/faceless-youtube-niches - 5 faceless YouTube niches for AI voice workflows
- /blog/podcast-network-project-management - Building a Podcast Network: Managing Multiple Shows with Projects
- /blog/case-study-50-youtube-shorts-week - How a Solo Creator Produces 50 YouTube Shorts a Week
- /blog/multilingual-content-scale - Multilingual Content at Scale: Reaching Global Audiences with 30 Languages
- /blog/elearning-producers-toolkit - The E-Learning Producer's Toolkit: Professional Course Audio Without the Studio
- /blog/voice-cloning-for-podcasters - Voice Cloning for Podcasters: Keep Your Voice, Skip the Mic
- /blog/valentines-day-audio-love-letter - Create a Personalized Audio Love Letter This Valentine's Day
- /blog/ai-bedtime-stories - Creating Magical Bedtime Stories with AI Voices
- /blog/ai-guided-meditation-audio - Creating Guided Meditations That Actually Sound Peaceful
- /blog/ai-voice-acting-indie-games - AI Voice Acting for Indie Game Developers: NPCs, Narration, and No Casting Calls
- /blog/
