Skip to content
Nappai Documentation

YouTube Transcripts

The YouTube Transcripts component lets you pull the spoken words from any public YouTube video. It can give you the whole transcript as one block of text or split it into timed chunks. You can also ask it to translate the transcript into another language.

How it Works

When you give the component a YouTube link, it first checks that the URL looks valid.
It then tries to grab the transcript using the official YouTube transcript API.
If that fails (for example, the video has no transcript or the API is blocked), it falls back to a second method that uses the pytube library to read any available captions.

Once the transcript is retrieved, the component can:

  • Text – combine all the spoken words into one long string.
  • Chunks – split the transcript into pieces that each cover a set number of seconds (default 60 s). Each chunk includes the text and a link that jumps to that part of the video.

If you choose a translation language, the component will translate the transcript into that language before returning it.

Inputs

  • Video URL: Enter the YouTube video link you want to transcribe.
  • Transcript Format: Pick text for a single block of text or chunks for timestamped sections.
  • Chunk Size (seconds): When you choose chunks, set how many seconds each chunk should cover (e.g., 60 s).
  • Language: Provide a comma‑separated list of language codes (e.g., en,es) in order of priority. Leave blank to use the default language.
  • Translation Language: Choose a language code (e.g., en, es, fr) to translate the transcript. Leave blank for no translation.

Outputs

  • Data – The raw transcript data.
    If you selected text, the output contains a single field called transcripts with the full transcript.
    If you selected chunks, the output is a list of objects, each with content (the text) and metadata (start/end times and a link to the video).

  • Tool – A reusable tool that can be called from other parts of your workflow. It performs the same transcription logic and returns the same data format.

Usage Example

  1. Add the component to your dashboard.
  2. Enter the YouTube URL of the video you want to transcribe.
  3. Choose the format you need:
    • Text – if you just want the whole transcript.
    • Chunks – if you need time‑stamped sections.
  4. If you chose Chunks, set the Chunk Size (seconds) (e.g., 60).
  5. Optionally, enter a Language list and/or a Translation Language.
  6. Run the component.
  7. The Data output will appear in the workflow. You can feed it into a summarizer, a translation tool, or store it in a database.
  • YouTube Search – Find videos by keyword before transcribing.
  • Text Summarizer – Condense the transcript into a short summary.
  • Translation – Translate the transcript into another language if you didn’t use the built‑in translation option.

Tips and Best Practices

  • Use the shortened YouTube URL (e.g., https://youtu.be/abc123) to avoid parsing errors.
  • If the video has multiple language tracks, list them in priority order (e.g., en,es,fr).
  • For long videos, consider using chunks to keep each piece manageable.
  • Check the Data output for an error field; if present, the component couldn’t retrieve a transcript.
  • Combine the transcript with a Text Summarizer to quickly get key points.

Security Considerations

  • Only transcribe videos you have permission to use.
  • The component does not store or share the video URL or transcript outside your workflow.
  • If you enable translation, the text is sent to the translation service used by the underlying libraries; ensure that service complies with your privacy requirements.