Craft your voice into unique characters with AI magic.
Tone Twist is a fun and innovative AI-powered application where users can upload or record their voice, select a character persona, and transform their speech into a completely new voice style.
Under the hood, the app combines cutting-edge technologies:
- Google Cloud Speech-to-Text for accurate transcription
- Gemini Flash 2.0 (Google AI) for enhancing the text according to the chosen character's speaking style
- ElevenLabs AI for ultra-realistic, expressive voice synthesis based on the persona
- 🎙 Upload or record your voice easily from the browser.
- 🎭 Choose your character (pirate, anime villain, robot assistant, etc.)
- ✨ Transform your voice through 3 stages: Transcription → Text Enhancement → Voice Synthesis
- 🔥 Live demo for each character with speaker previews.
- 🎧 Play and download your transformed voice instantly.
- 📈 Seamless progress tracking through interactive UI stages.
- Next.js 14 (App Router + Server Actions)
- Shadcn/UI for components
- TailwindCSS for styling
- Google Cloud Speech-to-Text API (Transcription)
- Gemini Flash 2.0 API (Text enhancement by persona)
- ElevenLabs Text-to-Speech API (Realistic voice generation)
- Vercel for hosting
-
Step 1: Add Your Voice
Upload an audio file (or record directly in-browser). -
Step 2: Choose Your Character
Select a pre-made persona that defines the emotion, accent, and style. -
Start Cookin'
The system:- Transcribes your speech into text
- Enhances the text based on the chosen character’s style
- Synthesizes natural-sounding audio using ElevenLabs voices
-
Enjoy Your Creation!
Listen, download, or share your newly twisted voice!
git clone https://github.com/MuditST/tonetwist.git
cd tone-twist
npm installCreate a .env.local file with:
GOOGLE_APPLICATION_CREDENTIALS=your-google-credentials.json
ELEVENLABS_API_KEY=your-elevenlabs-api-key
OPENAI_API_KEY=your-openai-or-gemini-api-keyThen run locally:
npm run dev- Maximum input audio length: 60 seconds (to avoid Google API timeout)
- Input audio is automatically downsampled to 16kHz mono WAV for best transcription performance
- User audio is processed securely and is not stored permanently
- Project optimized for modern browsers (Chrome, Edge, Safari)
- 🔒 User authentication (Clerk integration)
- 📦 Cloud history for saving previous transformations
- 🌐 Public sharing links for generated audios
- 🧠 Personalized AI voice personas in the future
This project is licensed under the MIT License.
- Built with ❤️ by Mudit Tushir
- Powered by OpenAI, Google Cloud, ElevenLabs, Vercel