OpenAI Official Audio Transcription API Guide

DuRT natively integrates OpenAI's official Audio Transcription endpoints. Known for exceptional multilingual accuracy and robustness in challenging acoustic environments, OpenAI models are ideal for complex academic interviews and commercial media projects.

---

Supported Model Matrix

Model	Word Timestamps	Auto Language Detect	Input Formats	Features & Notes
whisper-1	✅ Supported (Millisecond alignment)	✅ Supported	Local audio/video files & streaming audio	$0.006 / minute, gold standard for word-aligned subtitles
gpt-4o-transcribe	❌ Sentence-level	✅ Supported	Local audio/video files & streaming audio	Multimodal high-accuracy speech-to-text
gpt-4o-mini-transcribe	❌ Sentence-level	✅ Supported	Local audio/video files & streaming audio	Fast, cost-efficient speech-to-text

---

Getting Your OpenAI API Key

1. Log into OpenAI Platform Console;
2. Go to API Keys ➔ click "Create new secret key";
3. Copy your key (sk-proj-...).

---

Configuration in DuRT

1. Open DuRT, go to Settings ➔ AI Capabilities;
2. Under File Transcription, click "Add Service" ➔ select OpenAI ASR;
3. Fill in credentials:
   • API Key: Paste your key;
   • Base URL: Default is https://api.openai.com/v1;
4. Choose whisper-1 for millisecond-level subtitle timeline alignment, test connection, and save.