Audioscrape (Preview)

Search speech across podcasts, public meetings, and your own audio.

This connector is available in the following products and regions:

Service Class Regions
Copilot Studio Premium All Power Automate regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Logic Apps Standard All Logic Apps regions except the following:
     -   Azure Government regions
     -   Azure China regions
     -   US Department of Defense (DoD)
Power Apps Premium All Power Apps regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Power Automate Premium All Power Automate regions except the following:
     -   US Government (GCC)
     -   US Government (GCC High)
     -   China Cloud operated by 21Vianet
     -   US Department of Defense (DoD)
Contact
Name Audioscrape Inc.
URL https://www.audioscrape.com/support
Email support@audioscrape.com
Connector Metadata
Publisher Audioscrape Inc.
Website https://www.audioscrape.com/support
Privacy Policy https://www.audioscrape.com/legal/privacy
Categories AI;Data

Audioscrape

Audioscrape is the audio intelligence layer for AI agents: search spoken content across 250,000+ podcasts, government and civic meetings, and your own uploaded audio. Every result is grounded in speaker-identified transcript segments with timestamps, so agents can cite exactly who said what and when.

Publisher: Audioscrape Inc. Website: https://www.audioscrape.com Privacy policy: https://www.audioscrape.com/legal/privacy Terms of use: https://www.audioscrape.com/legal/terms Support: support@audioscrape.com · https://www.audioscrape.com/support Get started: https://www.audioscrape.com/docs/integrations/mcp Help: https://www.audioscrape.com/docs

Prerequisites

An active Audioscrape account is required. Sign-up is free at https://www.audioscrape.com/login and the core search experience works on the free plan. Paid plans (billed by Audioscrape) add private audio uploads, transcription requests, and higher usage limits.

Obtaining credentials

The MCP server uses OAuth 2.0 with PKCE. No API keys or license keys are used. When you add the server, you are redirected to Audioscrape to sign in and grant access; tokens are scoped to mcp:search, mcp:read, and profile.

  • Authorization URL: https://mcp.audioscrape.com/oauth/authorize
  • Token URL: https://mcp.audioscrape.com/oauth/token

Supported tools

Tool Description
search_audio Search all transcribed audio (public podcasts plus your own uploads) for a topic, phrase, or speaker; returns segments with timestamps.
search Simplified search interface returning citation-ready results.
fetch Fetch a single item by identifier for citation follow-up.
get_transcript Full speaker-identified transcript for an episode or upload, paged for long audio.
get_episode_overview Chapters and summary-level overview of one episode — the cheap first call before reading a full transcript.
list_recent_transcripts Recently transcribed items visible to the account.
list_podcast_episodes Episodes of a given podcast.
search_podcasts Find podcasts by name or topic.
search_speakers / get_speaker Find speakers and retrieve what they have said.
search_entities / get_entity_graph Entity search over the knowledge graph and the relationship graph around an entity.
get_trending / get_charts Trending topics and Apple/Spotify podcast charts.
transcribe_audio / get_transcription_status Request transcription of new audio and poll its progress.
list_my_datasets / list_my_uploads Enumerate the account's datasets and uploaded audio.
create_share_link Create a shareable link to a search or transcript view.

Getting started

  1. Add the Audioscrape MCP server to your Microsoft 365 Copilot, Copilot Studio, or Microsoft Teams agent (server URL: https://mcp.audioscrape.com).
  2. Sign in with your Audioscrape account when prompted.
  3. Ask your agent a question about spoken content — for example, "What did the mayor say about the transit budget in last week's council meeting?" or "Find podcast mentions of our company in the last month."

Known issues and limitations

  • Not every catalog item is transcribed. The catalog indexes far more episodes than are transcribed; search covers the transcribed corpus, which prioritizes in-demand shows. Use transcribe_audio to request specific items.
  • Long transcripts are paged. get_transcript returns a coverage object first; when truncated=true, call again with start_time=next_start_time. A truncated page means pagination, not missing transcription.
  • Some tools require a signed-in account with content: list_my_datasets, list_my_uploads, and transcribe_audio operate on the account's own workspace, and uploads require a paid plan.
  • Rate limits apply per account; heavy bulk retrieval should page through results rather than issue rapid parallel calls.
  • Transcripts are machine-generated (automatic speech recognition with speaker identification) and may contain recognition errors, especially for niche terminology or heavily accented or overlapping speech.

Actions

Audioscrape

Search speech across podcasts, public meetings, and your own audio.

Audioscrape

Operation ID:
InvokeMCP

Search speech across podcasts, public meetings, and your own audio.