The voice studio
that thinks
Neural TTS that runs entirely in your browser, guided by Claude's language intelligence. Script direction, multi-speaker casting, phoneme control — all at zero API cost.
“The neural network processes each phoneme with remarkable precision, generating natural-sounding speech in real time — entirely within your browser.”
How it works
Browser Synthesis
ClaudeSeiyuu Voice Engine runs entirely in your browser via WebAssembly and WebGPU. No audio ever leaves your device.
Claude Intelligence
OpenRouter routes your prompts to Claude for script direction, voice casting, and emotion tagging.
Zero Cost TTS
The voice model is free and open-source. Pay only for the Claude API calls you choose to make.
Voice Blending
Interpolate between two voice embeddings at runtime to create unique in-between voices on the fly.
Language G2P
Grapheme-to-phoneme conversion handles English with correct prosody and natural delivery.
Privacy First
All synthesis is local. History and presets are stored in localStorage only — no accounts, no tracking.
Everything you need
Script Director
Paste raw prose and Claude optimizes it for natural spoken delivery — pacing, pauses, rhythm — all corrected automatically.
Voice Blending
Interpolate two voices at any ratio. Create unique in-between characters that no voice catalog has.
Multi-Speaker
Assign different voices to different characters. Produce full cast recordings from a single script.
Phoneme Override
Fix mispronounced names using IPA notation. Precise control at the phoneme level.
Batch Pipeline
Queue multiple scripts and voices. Export all audio as WAV files once synthesis completes.
Real-Time Streaming
Hear each sentence as it synthesizes. No waiting for the full script — audio starts immediately.
Five ways to create
Every tool runs locally in your browser — no accounts, no upload limits.
Voice Studio
Write or paste any text, pick a voice, and synthesize speech instantly. The core ClaudeSeiyuu experience.
Open Studio →Voice Catalog
Browse and preview all 28 voices. Filter by language and gender — hear each one before you commit.
Browse Voices →Voice Blender
Interpolate between two voice characters with a ratio slider. Create unique in-between voices on demand.
Try Blending →Multi-Speaker
Paste a dialogue script. Claude assigns a unique voice to each character — full cast in seconds.
Cast Characters →Batch Pipeline
One line per entry. Process entire lists at once. Download each result as WAV.
Run Batch →vs Cloud TTS
| Feature | ClaudeSeiyuu | ElevenLabs | Google TTS | Azure TTS |
|---|---|---|---|---|
| Cost per 1M characters | $0 | $99+ | $16 | $15 |
| Runs in browser | ✓ | ✗ | ✗ | ✗ |
| Open source model | ✓ | ✗ | ✗ | ✗ |
| Voice blending | ✓ | ✓ | ✗ | ✗ |
| Multi-speaker script | ✓ | ✓ (paid) | ✗ | ✗ |
| Phoneme control | ✓ | ✓ | ✓ | ✓ |
| Privacy (no data sent) | ✓ | ✗ | ✗ | ✗ |
| Script intelligence (LLM) | ✓ | ✗ | ✗ | ✗ |
| Model download required | once, cached | ✗ | ✗ | ✗ |
9 languages, more coming
The studio interface
In the quiet village of Azabu, a young voice actress named Hana discovered an ancient manuscript…
Start creating
voice experiences
No sign-up. No API key for TTS. Open the studio and start speaking in any voice — right now.