🐍 PYTHON SDK

YouTube Transcript Python SDK — Official PyPI Package

Extract YouTube transcripts in your Python applications. Install from PyPI, authenticate with your API key, and integrate transcript extraction into your data pipelines, research tools, and automation scripts.

VIEW ON PYPI →API DOCUMENTATION

INSTALLATION

Terminal

pip install youtubetranscriptdevapi

QUICK START

Python

from youtubetranscript import YouTubeTranscript

yt = YouTubeTranscript("your_api_key")

result = yt.transcribe("dQw4w9WgXcQ")

print(f"Segments: {len(result.segments)}")
print(f"Duration: {result.duration:.0f}s")
print(f"Words: {result.word_count}")

for seg in result.segments[:5]:
    print(f"[{seg.start_formatted}] {seg.text}")

FEATURES

Python — Translation, Source, Format

# Translate to any language
result = yt.transcribe("dQw4w9WgXcQ", language="es")

# Choose caption source
result = yt.transcribe("dQw4w9WgXcQ", source="manual")

# Format options
result = yt.transcribe("dQw4w9WgXcQ", format={"timestamp": True, "words": True})

Python — Batch Processing (up to 100 videos)

batch = yt.batch(["video1", "video2", "video3"])
for t in batch.completed:
    print(f"{t.video_id}: {t.word_count} words")

Python — ASR Audio Transcription

job = yt.transcribe_asr("video_without_captions")
result = yt.wait_for_job(job.job_id)  # polls until complete
print(result.text)

Python — Export Formats

print(result.to_srt())               # SRT subtitles
print(result.to_vtt())               # WebVTT subtitles
print(result.to_plain_text())        # Plain text
print(result.to_timestamped_text())  # Text with timestamps

Python — Search & Account

# Search within transcript
matches = result.search("keyword")

# Account stats
stats = yt.stats()
print(f"Credits: {stats.credits_remaining}")

# History
history = yt.list_transcripts(search="python tutorial", limit=5)

ASYNC CLIENT

Python — asyncio

import asyncio
from youtubetranscript import AsyncYouTubeTranscript

async def main():
    async with AsyncYouTubeTranscript("your_api_key") as yt:
        # Single
        result = await yt.transcribe("dQw4w9WgXcQ")

        # Concurrent
        results = await asyncio.gather(
            yt.transcribe("video1"),
            yt.transcribe("video2"),
            yt.transcribe("video3"),
        )

asyncio.run(main())

ERROR HANDLING

Python

from youtubetranscript import YouTubeTranscript
from youtubetranscript.exceptions import (
    NoCaptionsError,
    AuthenticationError,
    InsufficientCreditsError,
    RateLimitError,
)

yt = YouTubeTranscript("your_api_key")

try:
    result = yt.transcribe("some_video")
except NoCaptionsError:
    job = yt.transcribe_asr("some_video")
    result = yt.wait_for_job(job.job_id)
except AuthenticationError:
    print("Check your API key")
except InsufficientCreditsError:
    print("Top up at youtubetranscript.dev/pricing")
except RateLimitError as e:
    print(f"Rate limited — retry after {e.retry_after}s")

API REFERENCE

Method	Description
transcribe(video, *, language, source, format)	Extract transcript
transcribe_asr(video, *, language, webhook_url)	ASR audio transcription
get_job(job_id)	Check ASR job status
wait_for_job(job_id, *, poll_interval, timeout)	Poll until ASR completes
batch(video_ids, *, language)	Batch extract (up to 100)
get_batch(batch_id)	Check batch status
list_transcripts(*, search, language, status, limit, page)	Browse history
get_transcript(video_id, *, language, source)	Get saved transcript
stats()	Account credits & usage
delete_transcript(*, video_id, ids)	Delete transcripts

TRANSCRIPT OBJECT

Property / Method	Description
segments	List of Segment objects
text	Full transcript as string
video_id	YouTube video ID
language	Transcript language
word_count	Total word count
duration	Total duration in seconds
to_srt()	Export as SRT
to_vtt()	Export as WebVTT
to_plain_text()	Plain text export
to_timestamped_text()	Text with [MM:SS] timestamps
search(query)	Find segments by text

SEGMENT OBJECT

Property	Description
text	Segment text
start	Start time (seconds)
end	End time (seconds)
duration	Duration (seconds)
start_formatted	"MM:SS" format
start_hms	"HH:MM:SS" format

CREDIT COSTS

Operation	Cost
Captions extraction	1 credit
Translation	1 credit per 2,500 chars
ASR audio transcription	1 credit per 90 seconds
Re-fetch owned transcript	Free

VS YOUTUBE-TRANSCRIPT-API (OPEN SOURCE)

You may have seen the popular open-source youtube-transcript-api package. Here is how ours differs:

Feature	YouTubeTranscript.dev SDK	youtube-transcript-api
AI transcription fallback	Yes — handles videos without captions	No — fails if no captions exist
Maintained & supported	Actively maintained with SLA	Community maintained, sporadic updates
Rate limit handling	Built-in with automatic retry	Manual — you handle rate limits
Playlist / bulk support	Built-in batch processing	Manual iteration required
YouTube blocking	Server-side — no IP blocking risk	Client-side — your IP can get blocked

OTHER SDKs & TOOLS

Language	Package
JavaScript / Node	youtube-transcript-api on npm
Dart / Flutter	youtubetranscript on pub.dev
MCP Server	mcp.youtubetranscript.dev

FREQUENTLY ASKED QUESTIONS

What Python versions are supported?+

Python 3.8 and above. We recommend Python 3.10+ for the best experience with type hints.

How is this different from youtube-transcript-api?+

Our SDK is the official package for YouTubeTranscript.dev. It includes AI-powered transcription for videos without captions, server-side extraction to avoid IP blocking, built-in rate limiting, and professional support. The open-source alternative only extracts existing captions and runs client-side.

Can I use it in Jupyter notebooks?+

Yes. The synchronous API works perfectly in Jupyter notebooks and Google Colab. Use the async API for concurrent processing in notebook environments.

Does it work with Django and Flask?+

Yes. The SDK is framework-agnostic and works with any Python web framework. Use the async interface with async frameworks like FastAPI.

Start Building with the Python SDK

pip install, authenticate, and extract transcripts in minutes.

VIEW ON PYPI →