Skip to content

willem4130/voice-noob

 
 

Repository files navigation

Voice Noob

Build AI voice agents that handle phone calls. No PhD required.

Built by Ken Kai · Join the community

Voice Noob


What Is This?

Voice Noob is an open-source platform for creating AI-powered phone agents. Configure an agent, connect a phone number, and let it handle calls — booking appointments, answering questions, or qualifying leads while you sleep.

The problem: Building voice AI is painful. You need to stitch together speech-to-text, LLMs, text-to-speech, telephony, and a dozen APIs. Then pray it doesn't sound like a robot.

The solution: Voice Noob handles the plumbing. Pick your AI tier, write a system prompt, enable tools, and deploy.


Why Voice Noob?

Feature Voice Noob Vapi Retell Bland AI
Open source Yes No No No
Self-hostable Yes No No No
Transparent pricing Per-minute breakdown Per-minute Per-minute Per-minute
Multi-LLM support 4 tiers (OpenAI, Google, Cerebras) Limited Limited OpenAI only
Built-in CRM Yes No No No
Tool integrations 30+ (Shopify, HubSpot, Calendly, etc.) Via custom code Limited Limited
Embeddable widget Yes Yes Yes No

TL;DR: Same capabilities as the paid platforms, but you own it.


Features

Dashboard

Voice Calling

  • Inbound and outbound calls via Telnyx or Twilio
  • Real-time transcription and call recording
  • ~320ms latency on Premium tier

AI Flexibility

Create Agent

  • Choose your price-to-quality tradeoff
  • OpenAI GPT-4 Realtime, Google Gemini, or Cerebras Llama
  • 13+ voice options

Built-in Tools

  • Contact management (search, create, update)
  • Appointment booking (check availability, schedule, reschedule)
  • Call controls (transfer, end call, DTMF)

30+ Integrations

Integrations

  • CRM: HubSpot, Salesforce, Pipedrive, Zoho, GoHighLevel
  • Calendars: Google Calendar, Outlook, Calendly, Cal.com
  • E-commerce: Shopify (orders, inventory, customers)
  • Communication: Slack, Gmail, SendGrid, SMS
  • Data: Airtable, Notion, Google Sheets
  • Payments: Stripe

Website Widget

  • Embed a voice agent on any website
  • One line of code
  • Domain allowlisting for security

Pricing Tiers

Pick based on your budget and quality needs:

Tier Cost/Min AI Provider Latency Best For
Premium $0.032 OpenAI GPT-4 Realtime ~320ms Quality-first use cases
Premium Mini $0.009 OpenAI GPT-4o Mini Realtime ~350ms High volume, good quality
Balanced $0.023 Google Gemini 2.5 Flash ~400ms Speed + cost balance
Budget $0.014 Cerebras Llama 3.3 70B ~530ms Maximum savings

Example: 1,000 calls/month × 5 min avg = 5,000 minutes

  • Premium: ~$160/month AI + $50 telephony = **$210/month**
  • Budget: ~$70/month AI + $50 telephony = **$120/month**

Quick Start

Prerequisites

  • Docker & Docker Compose
  • Python 3.12+
  • Node.js 20+
  • API keys for your chosen providers (OpenAI, Telnyx, etc.)

Setup

# Clone the repo
git clone https://github.com/KenKaiii/voice-noob.git
cd voice-noob

# Copy environment files
cp backend/.env.example backend/.env
cp frontend/.env.example frontend/.env

# Start PostgreSQL and Redis
docker-compose up -d

# Backend
cd backend
uv sync
uv run uvicorn app.main:app --reload

# Frontend (new terminal)
cd frontend
npm install
npm run dev

Open http://localhost:3000 and create your first agent.


Tech Stack

Voice & AI: Pipecat, OpenAI Realtime, Google Gemini, Deepgram, ElevenLabs Backend: FastAPI, PostgreSQL 17, Redis 7, SQLAlchemy 2.0 Frontend: Next.js 15, React 19, TypeScript, Tailwind, shadcn/ui Telephony: Telnyx (primary), Twilio (optional)


Project Structure

voice-noob/
├── backend/           # FastAPI Python backend
│   ├── app/api/       # API routes
│   ├── app/services/  # Business logic & integrations
│   └── app/models/    # Database models
├── frontend/          # Next.js frontend
│   ├── src/app/       # Pages (dashboard, agents, CRM)
│   └── src/lib/       # API clients, utilities
└── docker-compose.yml # PostgreSQL + Redis

Contributing

Pull requests welcome. For major changes, open an issue first.


License

AGPL-3.0

About

Voice agent platform for configuring and deploying custom AI voice agents

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • TypeScript 49.7%
  • Python 49.3%
  • Shell 0.4%
  • CSS 0.3%
  • JavaScript 0.2%
  • Makefile 0.1%