Natural voice for AI agents. Real-time WebSocket, OVOS bus, OpenAI-compatible gateway. Handle calls 24/7 with human-like voice.
65% of customers prefer phone calls. Text chatbots don't work for elderly people, the technologically illiterate, or hands-busy situations. And hiring a human call center costs $8-15 per hour in Ecuador.
Under 300ms end-to-end latency. Conversation flows naturally, without artificial pauses. Bidirectional audio streaming with automatic reconnection.
Integration with OpenVoiceOS for offline voice recognition. Works without internet. Ideal for rural areas of Ecuador with limited connectivity.
Compatible with OpenAI's voice API. Use GPT-4o for intelligent conversations, or connect your own fine-tuned model like Yachaq LLM EC.
Native Ecuadorian Spanish, Kichwa, and Quechua. Accents, idioms, and local intonations. Your agent sounds like an Ecuadorian, not a translated robot.
Django Channels for WebSocket. Whisper for STT. Coqui TTS for synthesis. OVOS for offline recognition.
Backend
Django + Channels
STT
OpenAI Whisper
TTS
Coqui XTTS
Offline
OVOS + Vosk
Protocol
WebSocket
Gateway
OpenAI-compatible
AgentVoiceVox is the spoken interface of the entire SomaTech ecosystem. Any agent can handle calls, answer queries, and close sales by phone.
Apache 2.0 License
Up to 1,000 voice minutes per month free. OpenAI gateway included. No per-channel costs.
On request
Unlimited minutes, Ecuador local numbers, integration with existing phone systems, and branded voice customization.
Give voice to your business with AgentVoiceVox. Your customers speak; your AI agents respond like humans.
No commitments. No credit card. Results in 48 hours.
