Audio & Video Transcriber

🎙️ Audio & Video Transcriber

Turn YouTube and TikTok videos, audio and video files, podcast feeds and Apple Podcasts episodes into text. Paste links and get the full transcript, timestamped segments, the detected language and ready-made SRT and VTT subtitle files, in the language spoken or translated into the language you choose. No login, no API key and no upload: the files are downloaded and transcribed on the fly with a Whisper speech model.

✨ Why use Audio & Video Transcriber?

• 🔗 Links in, text out: YouTube video links, TikTok video links, direct media files (MP3, M4A, WAV, OGG, OPUS, FLAC, AAC, MP4, MOV, WEBM...), podcast RSS feeds, Apple Podcasts show and episode links, and web pages that embed an audio or video file.
• 🌍 About 99 languages: the spoken language is detected automatically, or you set it. Optional translation into any language (for example Indonesian, Spanish or German), with the timestamps kept, also for speech that switches language.
• ⏱️ Timestamps and subtitles: segment start and end times, optional word timestamps, and SRT, VTT and TXT files saved for every file.
• 🎧 Built for podcasts: transcribe the newest episodes of any show, filter by date or episode title, and get the podcast and episode metadata with each transcript.
• 💸 Pay per audio minute: you pay for minutes actually transcribed plus a small start fee per run. Files that fail to download or contain no speech are free.
• 🔒 Private by design: speech recognition runs inside the run itself. Your audio is never sent to a third-party speech API. With 'Translate to', only the transcript text is sent to a language model to translate it.

🎯 Use cases

• 📰 Podcast monitoring: transcribe new episodes of competitor, industry or news podcasts and search them for brand mentions.
• 🧠 Content repurposing: turn episodes, webinars and talks into blog posts, show notes, quotes and newsletters.
• 🎬 Subtitles: create SRT/VTT captions for your own videos and audio.
• 🔎 Research and archives: make interviews, lectures, speeches and oral-history recordings searchable.
• 🤖 AI pipelines: feed clean transcripts with timestamps into LLMs, RAG systems and analytics.

🚀 How to use

1. Paste one or more links into Media files, podcast feeds or pages: a YouTube or TikTok video link, a direct file URL, a podcast RSS feed, an Apple Podcasts link or a web page with an embedded player.
2. For feeds, set Episodes per podcast feed (the newest episodes are taken first) and, if you like, the podcast filters.
3. Pick the Speech model (base is the balanced default) and, optionally, the spoken language and a language to translate into.
4. Click Start. Each file becomes one row in the dataset.
5. Export the results as JSON, CSV, Excel or HTML, or open the SRT/VTT/TXT links in each row.

📥 Input

{
  "urls": [
    "https://upload.wikimedia.org/wikipedia/commons/c/c4/GWBush_Oval_Office_Address_20010911-1-.ogg",
    "https://feeds.npr.org/500005/podcast.xml"
  ],
  "maxEpisodesPerFeed": 1,
  "model": "base"
}

📤 Output

[
  {
    "url": "https://feeds.npr.org/500005/podcast.xml",
    "mediaUrl": "https://prfx.byspotify.com/e/play.podtrac.com/npr-500005/npr.simplecastaudio.com/368f8510-314e-46f1-8d26-5b87ed6ab6eb/episodes/4b3cb9fd-8490-4628-80ef-bc730abac08e/audio/128/default.mp3?awCollectionId=368f8510-314e-46f1-8d26-5b87ed6ab6eb&awEpisodeId=4b3cb9fd-8490-4628-80ef-bc730abac08e&feed=O9WlY7a5&t=podcast&e=nx-s1-20261015-2000-long&p=500005&d=280&size=4586728",
    "sourceType": "feed",
    "title": "8pm ET, October 15 | The latest headlines from NPR",
    "podcastTitle": "NPR News Now",
    "podcastAuthor": "NPR",
    "feedUrl": "https://feeds.npr.org/500005/podcast.xml",
    "podcastImageUrl": "https://media.npr.org/assets/img/2023/01/24/npr-news-now_tile_sq-2334a5fd44563fa94cdcdb27384b0f05987c2d15.jpg",
    "episodeTitle": "8pm ET, October 15 | The latest headlines from NPR",
    "episodeGuid": "09ec1a41-8f71-44e0-914a-5ed2725ca733",
    "episodeUrl": null,
    "publishedAt": "2026-10-16T00:09:48Z",
    "feedDurationSeconds": 280,
    "mediaType": "audio/mpeg",
    "fileSizeBytesFeed": 4586728,
    "publisherTranscriptUrl": null,
    "fileName": "default.mp3",
    "fileSizeBytes": 5465268,
    "mediaFormat": "mp3",
    "audioCodec": "mp3float",
    "hasVideo": false,
    "durationSeconds": 341.58,
    "transcribedSeconds": 341.58,
    "truncated": false,
    "truncatedReason": null,
    "minutesLeft": 0,
    "status": "ok",
    "language": "en",
    "languageProbability": 0.9948,
    "task": "transcribe",
    "model": "base",
    "text": "This message comes from here there are blueberries, a secret album, mysterious photos, and an archivist's relentless search into the history trapped inside the lens. From the co-creators of the Laramie project, see this landmark new play strictly limited to 80 Broadway performances at the Barrymore Theatre. Previews begin November 2nd. Get tickets now at herethereourblueberries.com. Live from NPR  ...",
    "wordCount": 956,
    "segmentCount": 70,
    "segments": [
      {
        "start": 0,
        "end": 5.2,
        "text": "This message comes from here there are blueberries, a secret album, mysterious photos,"
      },
      {
        "start": 5.2,
        "end": 9.76,
        "text": "and an archivist's relentless search into the history trapped inside the lens."
      },
      {
        "start": 9.76,
        "end": 14,
        "text": "From the co-creators of the Laramie project, see this landmark new play"
      }
    ],
    "paragraphs": [
      {
        "start": 0,
        "end": 23.6,
        "text": "This message comes from here there are blueberries, a secret album, mysterious photos, and an archivist's relentless search into the history trapped inside the lens. From the co-creators of the Laramie project, see this landmark new play strictly limited to 80 Broadway performances at the Barrymore  ..."
      }
    ],
    "billedMinutes": 6,
    "processingSeconds": 143.01,
    "realTimeFactor": 0.419,
    "transcribedAt": "2026-10-16T01:07:00Z",
    "warning": null,
    "srtUrl": "https://api-staging.scrape.ly/v2/key-value-stores/<storeId>/records/8pm-ET-October-15-The-latest-headlines-from-NPR-0cd5a9ebd6fe.srt?signature=...",
    "vttUrl": "https://api-staging.scrape.ly/v2/key-value-stores/<storeId>/records/8pm-ET-October-15-The-latest-headlines-from-NPR-0cd5a9ebd6fe.vtt?signature=...",
    "txtUrl": "https://api-staging.scrape.ly/v2/key-value-stores/<storeId>/records/8pm-ET-October-15-The-latest-headlines-from-NPR-0cd5a9ebd6fe.txt?signature=..."
  }
]

❓ FAQ

Is it legal to transcribe this audio? The actor only downloads files that are publicly reachable. You are responsible for having the right to transcribe and use the content, for example your own recordings, public-domain material or content you have permission to process.

Do I need an account, API key or proxy? No. Downloads go through datacenter proxies by default. Switch to residential proxies only if a host refuses datacenter IPs.

Which model should I choose? base is fast and accurate on clear speech. small is noticeably better on noisy audio, strong accents and many non-English languages, but it takes about three times as long, so it uses more compute. tiny is the fastest and fine for quick drafts.

Can I get a transcript in another language than the one spoken? Yes: set translateTo, for example id for Indonesian. Every segment is translated and keeps its timestamps, so the SRT and VTT files are subtitles in that language; the original wording stays in originalText and segments[].originalText.

How long can a file be? Up to 300 minutes per file (default cap 120). Longer files are transcribed up to the cap and billed only for what was transcribed, with a warning in the row.

Why did a page not work? Many sites load their player with JavaScript or stream from a platform. In that case paste the direct file URL or the podcast's RSS feed.

Can I schedule it or use the API? Yes. Schedule a run with a podcast feed and publishedAfter: "1 day" to transcribe new episodes daily, or start runs and read the dataset through the Scrape.ly API, CLI, or from AI agents through the Scrape.ly MCP server.

⚠️ Limits

• YouTube video links work (the audio track is downloaded). Channel, playlist and search links, live streams, private and age-restricted videos do not.
• TikTok video links work: the video page link (https://www.tiktok.com/@user/video/...) or a share link (vm.tiktok.com/...). Profile, hashtag and search pages, photo posts and private or removed videos do not. TikTok's own media file links (the video/tos/... links a TikTok scraper returns in mediaUrls) expire and are served only with the TikTok session that created them, so paste the video page link (webVideoUrl) instead.
• Instagram, Facebook, X, Vimeo, SoundCloud, Spotify and other streaming platforms are not supported; use direct media files, podcast feeds and Apple Podcasts links.
• Speech that switches language is written in the language detected for each ~30-second piece; a very short phrase in another language inside a piece can come out in the piece's main language. A strong accent can be taken for another language: list the spoken languages (for example id,en) to rule that out. With 'Translate to' the whole transcript ends up in the chosen language.
• HLS streams (.m3u8) and DRM-protected media are not supported.
• No speaker labels (diarization) and no AI summaries.
• Whisper models can mishear names and jargon; the vocabulary hint helps. Accuracy depends on audio quality.
• Video files are downloaded in full, so they use more proxy traffic than audio files of the same length.