Turn Recordings Into Text and Text Into Speech Through Deepgram and Your Muse

Through Deepgram, your muse can turn a recording at a public link into a transcript with a summary, read text for topics and tone, and read a short passage aloud.

What it does

Deepgram is a speech service with three jobs. Connected through AnythingMCP, your muse gets 18 tools. First, it can transcribe a recording from a public link, with speaker labels, a summary, topics, intents, and sentiment for English audio. It can hide card numbers and personal data, detect the language, and use names you supply to get words right. Second, it can read text you already have and return a summary with topics, intents, and sentiment, no audio needed. Third, it can turn up to 2,000 characters of text into speech in an Aura voice and return an audio file you keep. The rest cover the account side: models, voices, projects, keys, usage, and credit.

How to use it

1. Start an AnythingMCP workspace and install the Deepgram connector.
2. Make a Deepgram account, open your project, then API Keys, and create a key. A Member key covers transcription, speech, and text reading. Enter it only where the connector asks for DEEPGRAM_API_KEY. The key is shown once, so save it with care.
3. In Muse, open Settings, then Connectors, and add a custom connector with the MCP server address AnythingMCP gives you. Sign in and approve.
4. Check the setup by asking the muse to list your projects, then give it a recording link and say what you want back.

What you need

• A Muse account that can add custom connectors
• A Deepgram account and API key, starting with free credit
• An AnythingMCP account with the Deepgram connector added
• A recording at a public link, or text to paste

What to watch for

• Local files cannot be uploaded through this connector. The recording must sit at a public link Deepgram can open; a link behind a sign-in will not work.
• The three working tools cost project credit: transcription by the minute, text reading by the text, and speech by the character. Ask the cost first, and save the audio file you get back, because Deepgram does not store it.
• Summaries, topics, and sentiment work on English only. Check names and numbers in a transcript before you share it.
• Every tool is on at install, including the ones that spend credit. This is a custom connector, and Meta does not review custom connectors. You can disconnect it in Settings at any time. The muse prepares, you decide.