OpenAI

OpenAI is used for AI-powered features like transcription correction.

Setup

1. Get API Key

1. Create an account at platform.openai.com
2. Go to API Keys
3. Create a new API key and copy it

2. Configure Environment

# .env or .env.local
OPENAI_API_KEY=sk-proj-...

Or configure it in your dashboard integrations.

Features

OpenAI is currently used for:

Transcription Correction

When using auto-generated captions with TTS audio, Whisper often misrecognizes brand names and technical terms. OpenAI's GPT-4o-mini is used to correct these errors:

import { compose, captions, audioModel } from "@synthome/sdk";

const execution = await compose(
  captions({
    video: "https://example.com/video.mp4",
    model: audioModel("vaibhavs10/incredibly-fast-whisper", "replicate"),
    originalText: "Synthome makes video editing easy.", // Enables correction
  }),
).execute();

Without originalText, "Synthome" might be transcribed as "Sintom" or "Sin Thome". With originalText, the transcription is corrected to match your original text while preserving word-level timestamps.

See Captions - Transcription Correction for more details.

Per-Request API Key

Override the environment variable for specific requests by configuring the key in your SDK initialization or dashboard.

Resources

• OpenAI Documentation
• API Keys
• Pricing