Transcribe Audio: Whisper + Make

An oral-history volunteer records hours of interviews with elders in the community. The recordings are precious and unusable: you cannot search audio, quote it, or share a passage without scrubbing through it. Transcribing by hand is roughly four hours of typing per hour of tape, so the tapes pile up untranscribed.

This build does the typing. Drop a recording into a Drive folder, and Whisper, OpenAI’s speech-to-text model, transcribes it into a Google Doc you can read, search, and quote. The interviews become text while you move on to the next one.

We wire it in Make, part of our meeting notes set, and it is the source step for the rest: once a recording is text, the recap and action-item builds can take over. This one uses Whisper because transcription is its job, not Claude’s.

What does this workflow actually do?

In one line: a recording becomes a document. Make watches a Drive folder. When you add an audio file, it sends the audio to Whisper, gets the transcript back, and saves it as a Google Doc. A folder of recordings becomes a folder of readable, searchable text.

A few real cases, none of them the usual ones:

• The oral-history volunteer above, finally turning hours of tape into quotable text.
• A journalist transcribing source interviews to find the right quote fast.
• A field researcher turning site recordings into searchable notes.
• A user-research lead transcribing usability sessions for the team to read.

Recording the conversation is easy; you press a button. Turning hours of audio into text you can use is the wall. Transcription is the tool built precisely to knock it down.

Why use Make instead of uploading to Whisper by hand?

Because a backlog is many files, and uploading each one, waiting, downloading the text, and filing it is the whole chore. Doing it once is fine; doing it for a shelf of recordings is not.

Make turns it into a pipeline. It watches the folder, sends every recording to Whisper, and saves each transcript as a Doc without a click. Whisper does the one hard step, turning speech into text; Make does the watching and filing. That division is what clears a backlog instead of starting one.

What do you need before you start?

• A Make account. The free 1,000 operations a month covers a lot of recordings.
• An OpenAI API key from the OpenAI platform (Whisper runs through it). New to API keys? See how to use Claude for the idea.
• A Google account with a Drive folder for recordings and somewhere to save the Docs.
• Audio in a common format (mp3, m4a, wav). Most phones and recorders produce these.

Two Make words. A scenario is the whole audio-to-text automation. A module is one box. Three boxes, one scenario.

How does the workflow work, step by step?

Three modules, left to right:

Module	App	What it does
1. Trigger	Google Drive	Fires when you add a recording to the folder
2. Brain	OpenAI Whisper	Transcribes the audio into text
3. Output	Google Docs	Saves the transcript as a document

Step 1: Watch a recordings folder

Create a scenario and add Google Drive, Watch Files in a Folder, connect your account, and pick the folder you will drop recordings into. This trigger fires on each new audio file.

Step 2: Transcribe with Whisper

Add OpenAI, Generate a transcription, paste your OpenAI key, and map the audio file from Step 1. Whisper handles the rest. If you know the language, set it; if your recording has long silences or music, trim it first for a cleaner result.

Tip: Whisper handles accents and background noise well, but a
clean recording always transcribes better. Good mic placement beats
any post-processing.

There is no prompt to write here: Whisper just listens and types.

Step 3: Save the transcript

Add Google Docs, Create a Document, and map Whisper’s text into a new doc named from the file. Now every recording has a readable, searchable companion. From here, point the recap or action-item build at this folder to go further.

How do you run it and check the result?

Click Run once after dropping a test recording in the folder. When I tested this, a few minutes of clear audio came back as accurate text in well under a minute, and the thing that mattered most was recording quality: a clean file read cleanly, a noisy one needed fixes. Run it, then turn the scenario on so recordings transcribe themselves.

After that, audio stops being a black box. Every recording becomes text you can search, quote, and feed into the rest of your meeting workflows.

What does this cost to run?

Piece	Free tier	If you outgrow it
Make	1,000 operations/month free	Core plan from about $9/month
OpenAI Whisper	Pay per minute	A few cents per audio minute, very cheap
Google Drive + Docs	Free	Free
Recorder	Free	Your phone works fine

Whisper bills by audio length and is cheap, cents per minute, so even long interviews cost little. The build sits inside the free Make plan. More on tiers in our Make guide.

What can go wrong, and how do you avoid it?

• Noisy audio transcribes poorly. Clean recording beats any cleanup; mind the mic.
• Big files time out. Very long recordings may need splitting into parts before transcription.
• Wrong language guessed. Set the language in the Whisper module if you know it.
• Sensitive recordings. Audio of private conversations is sensitive; keep the folder locked down and mind consent.

Once it is text, the rest of the set takes over. See Make AI scenarios for more.

How do you build this in Zapier or n8n instead?

Same three jobs, different names.

Job	Make	Zapier	n8n
Catch a recording	Watch Files in a Folder	New File in Folder trigger	Google Drive Trigger node
Transcribe it	OpenAI Generate a transcription	OpenAI Whisper action	OpenAI node
Save the text	Create a Document	Create Google Doc	Google Docs node

Make and Zapier are easiest to start. Zapier vs Make vs n8n compares all three.

Common questions about this build

Why Whisper and not Claude?

Transcription, turning speech into text, is a speech model’s job, and Whisper is built for it. Claude takes over once you have the text, to summarize or pull tasks.

How accurate is it?

Very, on clear audio, including accents and some background noise. Poor recordings are the main limit, so mind your microphone.

What formats does it take?

Common audio like mp3, m4a, and wav. Most phones and recorders produce these out of the box.

Can it transcribe video?

Extract the audio first, then transcribe that. Some tools do both; this build is for audio files.

Can I summarize the transcript too?

Yes. Point our recap build at the folder of transcripts, and Claude summarizes each one.

References and sources

• Creating a scenario (Make Help)
• OpenAI speech-to-text / Whisper (docs)
• Google Docs basics (Google)
• OpenAI on Make (integrations)
• Whisper — Grokipedia

Last reviewed: May 2026. Make, OpenAI, and Google update their interfaces; check the official pages for exact button names.

You may also like

• AI Automation for Beginners (the hub)
• Summarize new documents with Claude + Make
• Draft WordPress posts from a spreadsheet with Claude
• Make.com Complete Beginner’s Guide
• Make AI Scenarios: automations that save hours
• Zapier vs Make vs n8n: which to start with
• The AI Tools Directory