Vocalyne is a private, in-browser neural text-to-speech (TTS) and voice cloning web application. It runs artificial intelligence models directly on your device with WebAssembly and client-side inference, ensuring that no voice recordings, text prompts, or audio files are ever sent to a remote server.
Key Features
- 100% In-Browser & Private: All speech generation and voice cloning runs locally on your computer with zero server communication.
- Instant Voice Cloning: Clone any voice by recording 3 to 10 seconds of speech with your microphone or uploading an audio file (.wav, .mp3, .m4a).
- 20+ Expressive Neural Voices: Includes pre-tuned natural voices across diverse tones and styles.
- Offline Capable: After the initial one-time model setup, all files are stored permanently in browser cache for instant offline access.
- High-Quality Audio Export: Export generated speech as uncompressed, high-fidelity 16-bit PCM WAV audio files.
- Zero Subscriptions or Limits: Free and unrestricted speech synthesis with zero API rate limits.
Why I Built This
Most cloud text-to-speech services require paid API keys, enforce strict character caps, and upload your personal voice samples to third-party servers. Vocalyne proves that high-quality neural voice synthesis and voice cloning can run entirely on client hardware, giving users complete data privacy and unlimited access without recurring fees.
Tech Stack
- Frontend: Next.js, React, Tailwind CSS, TypeScript
- Audio & ML Engine: Client-side Neural TTS, Web Audio API, WebAssembly
- Storage & Caching: CacheStorage API for instant offline model loads
- Hosting: Vercel Edge Network