llama-index-readers-getyoutubetranscript

A LlamaIndex reader that loads YouTube video transcripts as Documents, powered by the GetYouTubeTranscript API.

It has the same load_data(ytlinks=[...]) call as YoutubeTranscriptReader, but transcripts are fetched by the API on its own servers. That means it keeps working on cloud servers, where YouTube blocks direct requests (RequestBlocked / IpBlocked), with no proxies to manage.

Install

pip install llama-index-readers-getyoutubetranscript

Get an API key at getyoutubetranscript.com/dashboard (free tier included):

export GETYOUTUBETRANSCRIPT_API_KEY=sk_live_...

Usage

from llama_index.readers.getyoutubetranscript import GetYouTubeTranscriptReader

reader = GetYouTubeTranscriptReader()
documents = reader.load_data(ytlinks=["https://youtu.be/jNQXAC9IVRw", "5e37ZT3SQbk"])

print(documents[0].metadata)
# {'url': 'https://www.youtube.com/watch?v=jNQXAC9IVRw', 'video_id': 'jNQXAC9IVRw', 'title': 'Me at the zoo',
#  'author_name': 'jawed', 'language_code': 'en', 'word_count': 39}

Switching from YoutubeTranscriptReader: change the import and class name; load_data(ytlinks=...) stays the same. Links can be any YouTube URL (watch, youtu.be, Shorts, live) or a video ID.

language and timestamps can also be passed per call: reader.load_data(ytlinks=[...], timestamps=True).

Build an index

from llama_index.core import VectorStoreIndex
from llama_index.readers.getyoutubetranscript import GetYouTubeTranscriptReader

documents = GetYouTubeTranscriptReader().load_data(ytlinks=["https://youtu.be/5e37ZT3SQbk"])
index = VectorStoreIndex.from_documents(documents)
print(index.as_query_engine().query("What does the speaker say about education?"))

Pricing

Each transcript uses one credit from your GetYouTubeTranscript account. Failed requests are not charged.

Links

• GetYouTubeTranscript API docs
• Python SDK (this package is built on it)
• License: MIT