Datasets:
metadata
language:
- en
license: other
task_categories:
- automatic-speech-recognition
tags:
- podcast
- transcription
- youtube
- rss
- diarization
size_categories:
- 10K<n<100K
configs:
- config_name: shows
data_files:
- split: train
path: shows/train.parquet
- config_name: episodes
data_files:
- split: train
path: episodes/train-*
- config_name: transcripts
data_files:
- split: train
path: transcripts/train-*
extra_gated_prompt: By requesting access you agree to follow our data use policies.
extra_gated_fields:
First Name: text
Last Name: text
I am a:
type: select
options:
- Projxon AI Member
- Non-Projxon AI Member
I want to use this dataset for:
type: select
options:
- Research
- Model Training (pre-training)
- Model Training (instruct)
- Other
Podcast Transcripts
Speaker-diarized transcripts from ~100k+ podcast episodes and YouTube videos (~45k hours of audio).
This dataset will be gated. Only people who are part of our team may access.
Splits
| Config | Rows | Description |
|---|---|---|
shows |
124 | Channels / podcast feeds (name, description, hosts, links) |
episodes |
102,374 | Episode/video metadata (title, description, guests, tags, dates) |
transcripts |
102,374 | ASR transcripts with diarized segments + YouTube official captions where available (~52%) |
Load
from datasets import load_dataset
ds = load_dataset("hudsongouge/Podcast-Transcripts-Raw")
shows = ds["shows"]
episodes = ds["episodes"]
transcripts = ds["transcripts"]
Or load configs individually:
episodes = load_dataset("hudsongouge/Podcast-Transcripts-Raw", "episodes", split="train")
Schema notes
Nested fields (hosts, guests, segments, official_transcript, metadata, etc.) are stored as JSON strings in Parquet. Parse with json.loads(row["metadata"]).
official_transcript (when present) contains publisher-provided captions — YouTube manual/auto subs with timed segments and plain text.
Links
episodes.episode_id→transcripts.episode_idepisodes.show_id→shows.show_id