Voice.ai combines a consumer voice changer with a developer platform for text-to-speech and phone-based voice agents. The TTS API returns synthesized speech either as a single response or as a stream, and the voice-agent product answers inbound calls and places outbound calls for a business, connecting to systems such as Salesforce, HubSpot, Zendesk, and Slack.
Streaming has three shapes: an HTTP chunked endpoint (/api/v1/tts/speech/stream) that delivers audio as it is generated, a single-context WebSocket that closes after one generation, and a multi-context WebSocket that runs several concurrent streams over one connection, each with its own context_id, voice, and settings. Requests name a model such as voiceai-tts-v1-latest or the lighter voiceai-tts-lite-v1-latest and authenticate with an API key in the Authorization header.
Features
- Streaming TTS: HTTP chunked and WebSocket delivery to cut time to first audio
- Multi-context sockets: concurrent streams with independent voices on one persistent connection
- Audio formats: MP3 at 22.05–44.1 kHz, Opus at 48 kHz, PCM from 8 to 48 kHz, and telephony A-law and mu-law at 8 kHz
- Voice agents: AI phone agents for inbound and outbound calls with CRM and helpdesk integrations
- Deployment: cloud or on-premise operation
- Voice changer: real-time voice conversion for consumer use
- Compliance: GDPR, SOC 2, and HIPAA
