Prompt
I'm building a support dashboard and need real-time transcription for calls. What API should I use?
Latest observation
Sep 27, 2026
OpenAI APIWeb search: off
For real-time call transcription in a support dashboard, the best fit is usually the OpenAI Realtime API.
Why use it
- Low-latency streaming transcription as audio arrives
- Works well for live support calls
- Can also power live summaries, sentiment, and agent assist in the same pipeline
- Supports audio input over WebRTC or WebSocket, which is useful for browser-based dashboards
Typical setup
- Capture call audio from your VoIP/contact-center stack
- Stream it to the Realtime API
- Receive partial and final transcripts in real time
- Show transcripts in your dashboard and optionally trigger actions like:
- keyword alerts
- QA notes
- escalation detection
- suggested replies
If you only need speech-to-text
If you don’t need live interaction and just want transcription from recorded audio, use the Speech-to-Text / Transcription API instead.
Practical recommendation
- Live calls + dashboard: Realtime API
- Recorded calls / batch jobs: Transcription API
If you want, I can also sketch a reference architecture for a support dashboard or give you a sample WebRTC/WebSocket implementation.