Scaled Cognition launches low-cost Voice TTS model for customer service

Scaled Cognition has introduced Voice, a text-to-speech model for enterprise customer service that accurately reads account numbers, addresses and codes in over 25 languages with voice cloning. It costs $9 per million characters via API or $1 per million for self-hosting, excluding infrastructure.

• API pricing is $9 per million characters; self-hosting is $1 per million
• Speech-generation latency is about 70 milliseconds
• Supports more than 25 languages and voice cloning
• Smart Pronunciation interprets meaning before generating audio