Devnagri

Devnagri specializes in real-time speech recognition for Indic languages. This bridge wires Teler call audio to Devnagri’s WebSocket; transcripts flow back into your application.

Reference repos

Python teler-devnagri-bridge

Node teler-devnagri-node-bridge

Prerequisites

A Devnagri account with a real-time WebSocket endpoint
A Teler account with an active API key
One Teler phone number on your account
An ngrok auth token (for local development)
Docker and Docker Compose

Configuration

Variable	Required	Default	Description
`DEVNAGRI_WS_URL`	yes	none	The Devnagri WebSocket URL provided by your Devnagri account.
`TELER_API_KEY`	yes	none	Your Teler API key.
`NGROK_AUTHTOKEN`	yes (local)	none	ngrok auth token.

git clone https://github.com/frejun-tech/teler-devnagri-bridge
cd teler-devnagri-bridge
cp .env.example .env
docker compose up -d --build

git clone https://github.com/frejun-tech/teler-devnagri-node-bridge
cd teler-devnagri-node-bridge
cp .env.example .env
docker compose up -d --build

Initiate a call

curl -X POST http://localhost:8000/api/v1/calls/initiate-call \
  -H "Content-Type: application/json" \
  -d '{
    "from_number": "+91XXXXXXXXXX",
    "to_number":   "+91XXXXXXXXXX"
  }'

When to choose Devnagri

Hindi, Bengali, Tamil, Telugu, Marathi, Punjabi, and other Indic languages.
Code-mixed speech (Hindi-English, etc.) where general-purpose STT models drop accuracy.
Cost-sensitive deployments in Indian markets.

For English-first agents, Deepgram is usually the better fit.

Troubleshooting

Symptom	Likely cause	Fix
Empty transcripts	Wrong sample rate	Confirm your Devnagri model expects 8 kHz
Connection refused	Stale `DEVNAGRI_WS_URL`	Refresh the WebSocket URL from Devnagri
Bridge doesn’t speak back	STT only by design	Add LLM + TTS in your fork

Next steps

Other recipes OpenAI Realtime, ElevenLabs, Gemini, and more.

WebSocket Messages Sending TTS audio back to Teler from your fork.