Wisprs

What Wisprs does

Wisprs provides hosted speech-to-text capabilities through the Model Context Protocol. It accepts audio and video sources, including YouTube and podcast links, and produces transcripts in more than 100 languages. The resulting transcripts can be searched, read in pages, summarized, translated, or transformed into other content.

The service is suited to agents that need to work with recorded conversations, published media, or an existing transcript library. It also supports exporting transcript content as TXT, SRT, VTT, DOCX, or Markdown files.

How it works

Wisprs is accessed as a hosted MCP server using Streamable HTTP. Client authentication uses OAuth 2.1, so a connected client can sign in without an API key. Once connected, an agent can submit media for transcription and use the available tools to inspect the resulting work.

Transcription may be handled as a job: agents can check its status before reading the transcript. Longer transcripts can be retrieved in pages rather than loaded all at once. Search tools let an agent find material in the transcript library, while generation tools operate on transcript content to produce summaries, chapters, translations, show notes, or a blog draft.

Setup and configuration

Use the Wisprs MCP setup endpoint to connect a compatible client over Streamable HTTP. The authentication flow is OAuth 2.1, and no API key is required. The service can be connected to Claude, ChatGPT, Cursor, Codex, or another client that supports Streamable HTTP.

After authentication, the client can discover and call the server’s MCP tools. A separate environment-variable configuration is not required for the hosted connection described here.

Tools and capabilities

Wisprs exposes tools for:

• Transcribing audio and video sources.
• Transcribing YouTube and podcast links.
• Selecting from more than 100 supported languages.
• Checking the status of transcription jobs.
• Reading transcripts page by page.
• Searching a transcript library.
• Generating summaries, chapters, and translations.
• Turning transcripts into show notes or a blog draft.
• Exporting content as TXT, SRT, VTT, DOCX, or Markdown.

These capabilities cover both the initial transcription step and common follow-up work. For example, an agent can transcribe a recording, locate relevant passages, create a summary, and export subtitle or document output within the same MCP connection.

Limitations and notes

Wisprs requires a client with Streamable HTTP support and an OAuth 2.1 sign-in. It is a hosted service rather than a local transcription process, and its workflow depends on submitting supported media or links to the service. The available output formats include plain text, subtitle formats, DOCX, and Markdown; selection depends on the export task the agent performs.