Spoken provides a REST API that fetches published podcast transcripts as Markdown with real speaker names and timestamps.
- for
- Developers building searchable podcast archives, LLM pipelines, or AI agents that need podcast text.
- pricing
- paid · free trial
- license
- MIT
Key features
- Episode listing — Free call returns metadata for every transcript-available episode of a show.
- Markdown transcripts — Returns full episode transcript in clean Markdown with bold speaker names and timestamps.
- Speaker name resolution — Identifies real host/guest names from context, avoiding generic diarization labels.
- Search by text or URL — Find episodes by title, guest, or by pasting Spotify, YouTube, or Apple podcast URLs.
- Agent integration — OpenAPI spec, MCP server, and npx skill let AI agents fetch transcripts directly.
- Credit-based pricing — Pay per episode (0.08-0.15 $) with credits that never expire and no subscription.
Use cases
- Build a searchable podcast knowledge base for LLM queries
- Create a podcast summarizer or RAG pipeline
- Keep an automated archive of new episodes for personal notes
- Power AI agents that answer questions about podcast content
Spoken vs alternatives
 Spoken |  Voice Calling SDK by Dyte |  gridctl |  Unified MCP Server |  A Workspace | |
|---|---|---|---|---|---|
| Best for | Podcast transcript retrieval API | Published podcast transcript retrieval | General MCP hosting | MCP integration platform | AI-enhanced coding workspace |
| Pricing | Paid | Subscription | Free | Subscription | Free |
| DevHunt upvotes | 1 | 44 | 20 | 29 | 6 |
| Launched | Jul 2029 | May 2024 | Aug 2026 | Jan 2026 | Jul 2026 |
- Spoken vs Voice Calling SDK by Dyte: Transcription APIs like Deepgram transcribe raw audio; Spoken fetches existing published transcripts.
- Spoken vs gridctl: Provides a unified MCP server for many tools, while Spoken is a specialized podcast transcript API.
- Spoken vs Unified MCP Server: Offers a broad MCP ecosystem; Spoken focuses solely on podcast transcripts with speaker names.
- Spoken vs A Workspace: Workspace is an AI-powered IDE; Spoken is an API for fetching podcast text, not a development environment.
Spoken FAQ
Is my show covered?+
Spoken only serves episodes that already have a published transcript; you can search for free to see availability.
How are speaker names determined?+
Spoken extracts names from the transcript context; if a name can’t be inferred it falls back to generic labels.
What if the service isn’t what I expected?+
You can request a full refund within 14 days if you used 20 credits or fewer; credits never expire.
Is there a free trial?+
Yes – a demo API key works for search and one full transcript without a card, plus two free trial credits.
Can I use Spoken with my AI agent?+
Yes – install the skill via npx, use the OpenAPI spec, or connect to the MCP server; any HTTP-capable agent can call the API.
What format are the transcripts returned in?+
Clean Markdown with bold speaker names and timestamps per turn, suitable for LLM context windows.
Summarized by DevHunt from spoken.md · Oct 6, 2026. Details may change; check the official site.
