Integrate automated hardcoded subtitle OCR, AI speech-to-text transcription, and subtitle format conversion directly into your localization, archiving, or AI training workflows.
Extract burned-in open captions from MP4, MKV, or MOV videos. Specify custom ROI crop boxes to isolate bottom subtitles from on-screen watermarks.
Generate time-synced subtitles from spoken audio tracks across 40+ languages with automatic language detection and clean line segmentation.
Download standard SRT/WebVTT subtitle files or consume structured JSON arrays with millisecond start/end timestamps and confidence scores.
# 1. Submit a video subtitle extraction job
curl -X POST "https://www.subextractor.com/api/v1/jobs" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"mode": "hardsub_ocr",
"videoUrl": "https://cdn.example.com/episode-01.mp4",
"language": "en",
"cropRegion": { "x": 0.05, "y": 0.78, "width": 0.90, "height": 0.18 },
"outputFormat": "srt"
}'
# 2. Example JSON response when completed
{
"jobId": "job_9f82a1c4",
"status": "completed",
"srtUrl": "https://www.subextractor.com/api/v1/jobs/job_9f82a1c4/download?format=srt",
"segments": [
{ "index": 1, "startMs": 1240, "endMs": 3890, "text": "Welcome back to the series." },
{ "index": 2, "startMs": 4120, "endMs": 6950, "text": "Today we are testing automated OCR." }
]
}Create and revoke API keys any time from your account dashboard.
Process batch video libraries in parallel without local GPU bottlenecks.
Pay only for processed video duration with pre-allocated quota protection.
The SubExtractor API allows developers and media teams to programmatically submit videos for hardcoded subtitle OCR extraction (with optional bounding box coordinates) or AI speech-to-text transcription, poll job status, and retrieve structured SRT, VTT, or JSON subtitle segments.
Sign in to your SubExtractor account, navigate to Account -> API Keys (/account/api), and generate a bearer token. API requests use your account's credit balance with transparent per-minute billing.
Yes. When creating an OCR extraction job, you can pass normalized bounding box coordinates (x, y, width, height) so the OCR engine only extracts captions from your target subtitle area and ignores watermarks or top banners.