Every new podcast episode and transcript, as files for your warehouse
Each day’s new episodes, transcripts and shows, ready for you to download from the app or via the API.
Enriched with every episode’s sponsors, hosts, guests and topics.
How it works
-
1
A day closes
Every episode transcribed or enriched that UTC day, every transcript stored that day, and every show Podzone follows.
-
2
It’s built the next morning
Three datasets, each as Parquet and as JSONL.gz, with a manifest of every file’s rows, size and checksum.
-
3
You download it
From the app’s Exports page, or list the days with one API request and fetch each file from a signed link. Downloads count against nothing.
-
4
You load it
Into DuckDB, BigQuery, pandas or your own pipeline, keeping the latest row for each episode.
Three datasets, a day at a time
Download only what you need: the episodes are small, and the transcripts are the bulk of each day.
Episodes
Transcribed or enriched that day
Each episode as the API returns it, with its show, and its sponsors with each ad read’s times and offer code, its hosts and guests, and what it’s about.
Transcripts
Stored that day, with timestamps
Every transcript in full, as timed segments, keyed by the episode’s id.
Podcasts
Every show, as of that day
Every show Podzone follows, with its author, categories, hosts and feed: a fresh snapshot each day.
Loads in one command
Parquet carries its column types: timestamps in UTC, and sponsors, people, topics and transcript segments as nested lists. Point a warehouse at the files and they load with nothing to declare first.
JSONL.gz holds the same rows, one JSON object a line, for tools that read JSON. Here is DuckDB, turning every day you’ve downloaded into one table:
-- The latest row for each episode, from every day downloaded
CREATE OR REPLACE TABLE episodes AS
SELECT *
FROM read_parquet('exports/episodes/*/*.parquet', hive_partitioning = true)
QUALIFY row_number() OVER (PARTITION BY id ORDER BY date DESC) = 1;
Checksummed
Every file’s sha256 is in the listing and in the day’s manifest, so a download cut short is caught before it’s loaded.
Complete, or not listed
A day appears only once all of its files and its manifest are in place, so you never fetch half a day.
Files that stay put
A day’s files don’t change, unless something is removed from Podzone: the day is then rebuilt without it, and listed with a later build time.
Download from the app or the API
Your account’s admins find every day in the app’s Exports page, each with its files and its manifest.
Or fetch them on a schedule: one request lists the days with a signed link to each file, and a short script downloads the ones you don’t have yet and checks them.
Example export day
Sun 15 Nov 2026
ManifestBuilt 16 Nov at 04:12.
| Dataset | Parquet | JSONL.gz |
|---|---|---|
| EpisodesTranscribed or enriched that day | Download | Download |
| TranscriptsStored that day, with timestamps | Download | Download |
| PodcastsEvery show, as of that day | Download | Download |
Exports, the firehose or the API?
Exports
Everything new, a day at a time, transcripts included: to fill your own warehouse, run your own analysis or keep your own copy.
The firehose
New episodes from the shows you choose, pushed to your server as they’re transcribed, with each transcript a link away. More on the firehose.
The API
One question at a time: search what was said, read any episode’s transcript, and fill in what came before your first file. More on the API.
Questions
Which plans include the export files?
Business. See the plans for how many days it keeps.
Are full transcripts included?
Yes: every transcript stored that day, in full, as timed segments. Downloading the files uses none of your plan’s transcript downloads; listing the days through the API is one request.
How far back do the files go?
Each day is kept for your plan’s window and then deleted, so download the days you want to keep. For older episodes, use the API. Longer history is part of Enterprise, on the pricing page.
Can I choose which shows are in the files?
No: the files hold everything new, the same for every customer, and you keep what you need once they’re loaded. For the shows you choose as they arrive, use the firehose; for the episodes that mention you, use alerts.
Why is the same episode in more than one day?
A day holds what changed that day: an episode is in the day it was transcribed, and again in the day its sponsors, people and topics were completed. Keep the latest row for each episode, or upsert on its id as each day arrives.
What may I do with the files?
Use them in your business and in the products you build for your customers. Keep them secure and don’t pass them on; redistributing them or training models on them is part of Enterprise. The Terms of Service say exactly what is allowed.
For developers
- Bulk exports Download each day's new episodes, transcripts and shows as Parquet and JSONL.gz, and load them with DuckDB, BigQuery, pandas or your own tools.
- Quickstart From an API key to your first search, your first alert and your first signed webhook delivery.
- API reference Every endpoint, field and error.
Related
Start free, and upgrade when you need more.
Every plan includes alerts, search, transcripts, sponsors, people, topics, the API and the MCP server for AI agents. See the plans.
Start free