Voice Sample Audio Files

Pre-recorded audio samples for the landing page "Six voices. One conversation." demo.

Quick Start

Generate all voice samples using the Ferni CLI:

ferni voices generate-samples

This uses:

• Persona voice registry - Same voice IDs as the production app
• Cartesia sonic-3 - Same TTS model as PersonaAwareTTS
• Persona-specific voices - Ferni, Maya, Peter, Alex, Jordan, Nayan

Samples Generated

Requirements

• CARTESIA_API_KEY in your .env file
• Voice IDs configured in src/config/voice-ids.ts

Audio Format

• Format: MP3
• Sample rate: 44.1kHz
• Model: Cartesia sonic-3 (same as production)

Deployment

After generating samples:

npm run deploy:landing

The samples are served from /audio/samples/ on the landing page.

How It Works

The voice-samples.js module on the landing page:

1. Tries to load pre-recorded MP3 files first
2. Falls back to browser TTS if files aren't found
3. Shows persona name, role, and color for each sample

Updating Samples

To update the voice samples:

1. Edit responses in scripts/ferni.ts → handleVoices → generate-samples
2. Regenerate: ferni voices generate-samples
3. Deploy: npm run deploy:landing

The sample text should match apps/website/ferni-website/src/js/voice-samples.js.