Teler WAV Bridge

Overview

A reference implementation demonstrating how to stream a pre-recorded WAV file to a live phone call using Teler’s Media Streaming Bridge over WebSockets, while recording and collecting the caller’s responses.

This project demonstrates a simple but powerful voice workflow: initiate an outbound phone call, stream a pre-recorded WAV file to the callee over WebSockets, capture the callee’s responses in real time, and store the call recording for later review and analysis.

Unlike AI voice agents, there’s no real-time speech generation or conversational AI involved. The caller hears a pre-recorded audio message, and their responses are recorded and made available after the call. This makes the integration lightweight, predictable, and ideal for one-way announcements, surveys, campaigns, and response collection workflows.

Who is this for? Teams that want to automate voice outreach without deploying AI agents.

Key Features

• Pre-recorded audio, no AI pipeline — Stream a WAV file directly into a live call over WebSockets, with no STT, LLM, or TTS involved.
• Captures caller responses — Records whatever the caller says back, available for review after the call.
• Dockerized deployment — Clone the repo and run it with a single Docker Compose command.
• Recordings accessible two ways — View and manage recordings from the FreJun Teler Dashboard, or retrieve them programmatically through the Recording API.

Common Use Cases

Appointment reminders

Play a pre-recorded reminder and collect customer confirmations or responses.

Recruitment and candidate screening

Play screening questions and collect recorded candidate answers.

Customer feedback campaigns

Gather voice feedback without requiring a live agent or AI assistant.

How to Integrate

Prerequisites

• Node.js 18+
• Docker and Docker Compose
• A Teler API key (get one at platform.frejun.ai)
• An ngrok account and auth token

Step 1: Clone and configure

git clone https://github.com/frejun-tech/teler-wav-bridge.git
cd teler-wav-bridge

cp .env.example .env

Update the .env file with your credentials.

Step 2: Run with Docker

docker compose up -d --build

Environment variables

Audio requirements

For best results, audio files should be:

• WAV format
• PCM S16LE
• Mono
• 8000 Hz sample rate

This format aligns with standard telephony audio and ensures optimal playback quality during calls.

API endpoints

Health & Status

• GET / – Health check and server information
• GET /health – Service status
• GET /ngrok-status – Current ngrok status and public URL

Calls

• POST /api/v1/calls/initiate-call – Initiate an outbound call
• POST /api/v1/calls/flow – Retrieve call flow configuration

Media Streaming

• WebSocket /api/v1/calls/media-stream – Streams pre-recorded WAV audio to the call and receives caller audio events.

Webhooks

• POST /api/v1/webhooks/receiver – Receives call status and media events from Teler.

Call initiation example

curl -X POST "https://your_ngrok_domain/api/v1/calls/initiate-call" \
  -H "Content-Type: application/json" \
  -d '{
    "fromNumber": "+918064xxx",
    "toNumber": "+919967xxx"
  }'

Integration at a Glance