Fonoster — Build AI Voice Apps Like Twilio, But Open Source
Why Fonoster
Twilio charges per minute and per API call. Fonoster is self-hosted and open-source — you control your telephony infrastructure and costs.
Programmable Voice Verbs
| Verb | Description |
|---|---|
| Answer | Pick up an incoming call |
| Say | Text-to-speech output |
| Gather | Collect user input (speech or DTMF) |
| Dial | Connect to another number |
| Record | Record the conversation |
| Play | Play an audio file |
Example: AI Customer Service Bot
const { VoiceServer } = require("@fonoster/voice");
new VoiceServer().listen(async (req, res) => {
await res.answer();
await res.say("Welcome to AI Support. How can I help?");
const input = await res.gather({
source: "speech",
timeout: 5000
});
// Process with AI (OpenAI, Claude, etc.)
const aiResponse = await processWithAI(input);
await res.say(aiResponse);
await res.hangup();
});Key Features
- NodeJS SDK — familiar JavaScript API for voice apps
- OAuth2 & JWT — secure authentication
- RBAC — role-based access control
- Multi-tenant — serve multiple organizations
- Google Speech — built-in speech recognition
- S3 Storage — store recordings and assets
- SSL/TLS — Let's Encrypt support
- Docker — easy deployment
FAQ
Q: What is Fonoster? A: An open-source telecom platform (Twilio alternative) for building AI voice applications with programmable voice verbs, speech recognition, and a NodeJS SDK.
Q: Is Fonoster free? A: Yes, fully open-source under MIT. Self-host on your own infrastructure.
Q: Can Fonoster integrate with AI models? A: Yes, use the Gather verb to capture speech, process it with any AI model (OpenAI, Claude, etc.), and respond with Say.