Subtitles
The original caption track, with language and timing, when YouTube actually has one.
YouTube subtitles · video transcription · AI summary
AI Video Subtitle & Transcription Generator
Extract YouTube subtitles, transcribe video and audio, and summarize with AI. Then get chapters, highlights, action items, and chat grounded in the transcript.
Start from any supported source
YouTube captions work now. Video and audio uploads are transcribed, then the original file is deleted.
Paste a public YouTube URL or a direct MP3/MP4 link. YouTube captions are extracted first.
Guests can process one item. Free accounts get 3 shared attempts per day. Uploads are temporary.
How it works
Captions first. AI second. You never wait on a fake transcript.
- 1
Paste a YouTube URL
We accept watch, Shorts, embed and youtu.be links and extract the video ID.
- 2
Extract captions
Manual tracks are preferred. Auto-generated captions are used when that is all YouTube provides.
- 3
Clean the transcript
Repeated fragments and caption noise are removed. Timestamps stay intact.
- 4
AI analysis
GPT-5 nano writes the executive summary, ~20% condensed content, chapters, highlights and chat.
What you get
Subtitles, transcription and summary stay distinct. Every timestamp can jump back into the video.
Full transcript
A cleaned video-to-text reading view — not the same thing as the AI summary.
AI video summary
1–2 paragraph executive summary plus condensed content targeting about 20% of the source.
Chat with timestamps
Ask a question and skip to the matching moment instead of rewatching the whole file.
Core features
An AI video summarizer and transcription tool — not a YouTube gadget with a summary bolted on.
AI video summary
AI Video Summarization
After a real transcript exists, GPT-5 nano writes an executive summary of 1–2 paragraphs and a condensed representation targeting about 20% of the original content. It is not a rewrite of the whole video, and it does not run until captions or a transcript have actually been obtained.
Video to text
Automated Video Transcription
YouTube videos are transcribed from available captions today. Uploaded MP4, MOV and WEBM files will use speech-to-text in Phase 2. Timestamps stay attached either way.
Audio to text
Automated Audio Transcription
Audio transcription for MP3, M4A and WAV is planned for Phase 2 with Whisper large-v3 or an equivalent configured provider. The product is designed for audio, not only YouTube.
YouTube subtitles
YouTube Subtitle & Caption Extraction
The live pipeline discovers caption tracks, prefers official/manual subtitles, supports auto-generated captions when present, and preserves languages and timestamps. If no usable captions exist, we say so. We never invent a transcript.
Timestamped chapters
Chapters & Highlights
Existing YouTube chapters are kept when they are suitable. Otherwise chapters are generated from the transcript. Each job also includes about 5–10 highlight moments with clickable timestamps.
Meeting intelligence
Action Items & Decisions
When the source contains them, action items are listed separately from decisions. Empty lists mean none were found — we do not fabricate tasks.
Grounded Q&A
Chat with your video
Ask questions against the transcript. Answers are retrieved from timestamped chunks so you can skip to the relevant moment instead of rereading the whole file.
Video subtitle generator
Hardcoded subtitle generation
From extracted captions or a transcript we can produce SRT, VTT and TXT subtitle files. On-screen viewing works now. Protected downloads require a registered, verified account in the production release.
Supported sources and use cases
YouTube transcription is live. Uploaded video and audio are specified for Phase 2. Direct media URLs may follow, with strict safety limits.
YouTube
Available nowWatch, youtu.be, Shorts, embed and mobile URLs. Caption-first extraction with multi-method failover.
Video files
Phase 2MP4, MOV and WEBM. Temporary processing only — original files are not kept as a media library.
Audio files
Phase 2MP3, M4A and WAV for meetings, podcasts and voice notes, transcribed with configured speech-to-text.
Launch use-case copy. Replace or append real customer stories in src/content/site.ts as usage grows.
Content creators
Turn YouTube videos into a transcript, AI video summary, chapters and highlight quotes you can reuse in descriptions, blogs and shorts.
Podcasters
Audio transcription for episodes is planned in Phase 2. Today you can summarize a published YouTube version of the show with captions.
Marketers
Extract a YouTube transcript, get key takeaways and timestamped highlights, then brief a campaign without watching the full recording.
Students
Paste a lecture URL to get YouTube subtitles, a full transcript and an AI video summarizer output you can study from.
Researchers
Keep timestamps attached to every cue so quotes stay grounded in the source video instead of a rewritten paraphrase.
Business professionals
Use chapters, action items and chat-with-video on public recordings now. Meeting file uploads follow in the production release.
Trusted by teams who live in video
Figures, logos, G2/Product Hunt scores and testimonials below are editable launch placeholders in src/content/site.ts. Swap them for genuine reviews when you have them.
- Videos processed
- 12,400+
- Caption languages
- 50+
- Word-level timestamps
- Yes
- Transcript accuracy target
- 95%+
G2 4.8 / 5
Placeholder rating — replace with live G2 reviews
Product Hunt 4.7 / 5
Placeholder rating — replace with live Product Hunt data
Northwave MediaLectureLabPulseCastHarbor ResearchStudioFieldBrightfunnel
I paste a YouTube lecture and get captions, a transcript, and a usable summary with chapters in one pass. It saves me from rewatching hour-long videos.
The distinction between existing YouTube subtitles and a generated transcript is clear. We use the timestamped highlights to clip social posts.
Action items from meeting recordings will be the reason we upgrade. Even the YouTube flow already feels like a serious transcription product.
Frequently asked questions
Straight answers about YouTube transcription, AI subtitles, free limits and what is still in Phase 2.
What is AI Video Summary?
AI Video Summary is a video and audio intelligence platform. It extracts YouTube subtitles and captions when they exist, produces a clean full transcript, and generates an AI video summary, takeaways, chapters, highlights, action items and chat grounded in that transcript.
Can I summarize a YouTube video?
Yes. Paste a public YouTube URL. The app first discovers existing captions. After a usable transcript is obtained, GPT-5 nano produces an executive summary (1–2 paragraphs) and condensed content targeting about 20% of the original length.
Can I upload a video?
Video upload is part of Phase 2. Supported planned types are MP4, MOV and WEBM. The homepage shows the upload option so you can see it is coming; it is not a working processor yet.
Can I upload an audio file?
Audio upload is also Phase 2. Planned types are MP3, M4A and WAV, with speech-to-text using Whisper large-v3 or an equivalent configured provider.
Does it extract existing YouTube captions?
Yes. YouTube subtitle and caption extraction is the highest-priority path. The system prefers official/manual captions, also supports auto-generated captions when available, and never claims subtitles exist if none were actually found.
What happens when a YouTube video has no captions?
The app reports that honestly. It does not invent a transcript. Speech-to-text fallback and file upload for captionless videos are planned where technically and legally appropriate. You can also upload the file once Phase 2 processing is live.
What video and audio formats are supported?
YouTube URLs work now (watch, youtu.be, Shorts, embed and mobile links). Uploaded video (MP4, MOV, WEBM) and audio (MP3, M4A, WAV) are specified for Phase 2. Direct media URLs may be added later with strict SSRF protections.
Can I generate subtitles?
When captions or a transcript exist, the product can produce subtitle representations such as SRT, VTT and TXT. Protected file downloads require a registered, verified account and ship with the full production release.
Can I get a full transcript?
Yes. After captions are extracted they are cleaned into a full transcript with clickable timestamps. That transcript is distinct from the AI summary.
Can I get timestamps and chapters?
Yes. Existing YouTube chapters are used when present; otherwise chapters are generated from the transcript. Highlights and chat answers also carry timestamps you can jump to in the video.
Can I chat with my video?
Yes, once a job is ready. Chat is grounded in the transcript with retrieved timestamped passages. It should not invent facts that are not in the source.
How accurate is the transcription?
When YouTube already provides captions, accuracy follows that source (manual tracks are preferred over auto-generated). For uploaded media, Phase 2 will use Whisper large-v3 or an equivalent high-quality speech-to-text service. We do not invent missing speech.
How does the free usage limit work?
Unregistered visitors may process one eligible item. After that you must register. Free registered users get 3 shared processing attempts per day across YouTube and, later, uploads. Limits are enforced on the server, not in the browser.
Do I need an account?
No account is required to try one YouTube video and view results on-screen. Registration is required for further processing. Protected downloads will require a verified email in the production release.
What happens when I reach my free limit?
You will see a professional message asking you to contact us to upgrade. There is no public pricing page and no self-service checkout.
Start with a YouTube URL
Extract subtitles, get a full transcript, and generate an AI video summary. Uploads arrive in Phase 2.