Auto Subtitles
Generate subtitles from speech using AI. Choose from 5 engines: Parakeet (lightning fast), Whisper (99 languages), or Qwen3 (52 languages). GPU-accelerated on server.
Drop files here or click to browse
Max 10240.0 MB per file · drop multiple for batch
Processed on our servers — requires a free account
Have feedback? Let us know
How to Use Auto Subtitles
- Upload a video with speech
- Select the language (or leave on auto-detect)
- Pick an AI engine — Parakeet for speed, Whisper for accuracy, Qwen3 for Asian languages
- Click Process and download your SRT subtitle file
Features
- 5 AI engines: Parakeet TDT (NVIDIA), Whisper Large V3 Turbo, Whisper Large V3, Whisper + Word Timestamps, Qwen3-ASR
- Up to 99 languages with auto-detection
- Parakeet Lightning: transcribe 10 minutes of audio in under a second
- Word-level timestamps for karaoke-style subtitles
- Exports SRT subtitle files
Frequently Asked Questions
- Which AI engine should I use?
- Parakeet TDT is the fastest and most accurate for English/European languages. Whisper Large V3 Turbo is a great default for any language. Whisper Large V3 provides the highest accuracy. Qwen3-ASR is best for Asian languages like Chinese, Japanese, and Korean.
- What languages are supported?
- Whisper supports 99 languages. Parakeet supports 25 European languages. Qwen3-ASR supports 52 languages including many Asian languages. Auto-detect works across all engines.
- What are word-level timestamps?
- The 'Whisper + Word Timestamps' option gives you precise timing for each individual word, useful for karaoke-style subtitles or precise caption sync. Other modes give sentence-level timing.
- How fast is Parakeet?
- Parakeet TDT processes at over 3,000x real-time — a 10-minute video takes less than a second of compute. It's the fastest open-source speech recognition model available.