To automatically transcribe meeting recordings, upload your audio or video file to an AI transcription tool such as Otter.ai, Descript or OpenAI’s Whisper, let it process the file, then review and export the text. The whole process usually takes a few minutes of your attention and a few minutes of machine time — no typing, no rewinding, no paying a human transcriptionist by the audio hour.

This guide walks through the setup for three different tools, explains which one suits which kind of worker, and shows how to turn a raw transcript into something your team actually uses. It is written for people who have never touched transcription software before, so nothing is assumed.

What automatic meeting transcription actually does

An AI transcription tool listens to your recording and converts speech into written text. Most modern tools do three extra things on top of that:

• Speaker labels (diarisation). The software attempts to separate “Speaker 1” from “Speaker 2” and lets you rename them.
• Timestamps. Every paragraph is tagged with the point in the recording where it was said, so you can jump straight to a moment.
• Summaries and action items. Many tools now add an AI-generated recap on top of the transcript.

What it does not do reliably is read minds. Accuracy drops with crosstalk, heavy background noise, strong accents the model has seen less of, and jargon or product names it has never encountered. Treat every transcript as a strong first draft that needs a skim, not as a legal record.

Step 1: Prepare your recording before you upload anything

Accuracy is decided before the AI ever sees your file. A clean recording can be the difference between a transcript you skim and a transcript you rewrite.

Record each speaker separately where you can

Zoom, Google Meet and Microsoft Teams can all save a cloud recording, and some plans allow separate audio tracks per participant. Separate tracks dramatically improve speaker labelling, because the software no longer has to guess who is talking.

Ask people to use headsets

Laptop speakers plus a laptop microphone creates echo, and echo confuses diarisation. Any cheap wired headset removes most of it.

Check your file format

MP3, M4A, WAV and MP4 are accepted almost everywhere. If your recorder produces something unusual, convert it first. Also check file size limits — free plans often cap both the length of a single file and the total minutes per month.

Decide where the file is allowed to go

Cloud transcription means your meeting audio leaves your machine. If you discuss client data, health information or anything under an NDA, check the provider’s privacy and data-retention policy, and consider a local tool instead. This is a genuine decision point, not a formality.

Three AI transcription tools compared

Each of these takes a different approach: one is built around live meetings, one around editing, one around running the model yourself. Plans and limits change frequently, so the table below describes the shape of each offering at time of writing — always confirm current details on the provider’s own pricing page before you commit.

Step-by-step setup for each tool

Option A: Otter.ai for live meeting capture

1. Create an account and open the web dashboard.
2. Connect your calendar (Google or Outlook). This is what lets the assistant see upcoming meetings.
3. In settings, choose whether the assistant auto-joins every calendar meeting or only ones you select. Start with “selected meetings” so you do not surprise anyone.
4. To transcribe Zoom meetings with AI, either let the assistant join as a participant, or record locally and upload the file afterwards.
5. For an existing recording, click Import and select your audio or video file. Processing is usually a fraction of the recording’s length.
6. Open the finished transcript, rename the speakers, and use the search box to find key moments.
7. Export as TXT, DOCX, PDF or SRT depending on your plan.

Tell people they are being recorded. In many places consent is a legal requirement, and in all places it is basic courtesy. A bot silently appearing in a call damages trust fast.

Option B: Descript for transcript-driven editing

1. Install the desktop app and create a new project.
2. Drag your meeting recording into the project. Descript transcribes it automatically on import.
3. Open the transcript view. Deleting a sentence of text also deletes that audio — this is the feature that makes Descript different.
4. Use speaker detection to split the transcript, then correct any mislabelled sections.
5. Trim filler words in bulk with the filler-word removal tool.
6. Export the transcript as text, or export a trimmed video clip for a highlight reel. The same workflow applies to any recorded video, which I cover in more depth in this beginner’s guide to transcribing videos with AI and generating auto subtitles.

Option C: Whisper for free, private, unlimited transcription

1. Install Python on your computer, then install Whisper using pip from a terminal window.
2. Install FFmpeg, which Whisper uses to read audio files.
3. Place your recording in an easy-to-find folder.
4. Run the transcription command, choosing a model size. Smaller models are fast and rougher; larger models are slower and more accurate. Try the medium size first.
5. Whisper writes out TXT, SRT and VTT files next to your audio.
6. Open the TXT file and clean it up. Note that base Whisper does not label speakers — you need an additional diarisation step for that.

Whisper runs much faster on a machine with a dedicated GPU. On an older laptop, a long meeting may take a long time to convert audio to text automatically, so start it running and go do something else.

Costs, and what free plans genuinely cannot do

I am not going to quote prices, because they change and because the only trustworthy source is the vendor’s own page. What is stable is the pattern of restriction:

• Monthly minute caps. Free tiers almost always limit total transcription minutes. A few long weekly team calls can exhaust them quickly.
• Per-file length caps. A free plan may refuse a two-hour workshop recording even if you have minutes left.
• Limited uploads. Some free plans allow only a handful of imported files, ever — live capture only after that.
• Export restrictions. Plain text may be free while DOCX, PDF, SRT or bulk export sit behind a paid plan.
• Retention windows. Older transcripts can be archived or deleted on free accounts. Download anything you need to keep.
• AI summaries and search. Chapter summaries, action-item extraction and cross-meeting search are common upsells.

Whisper is the honest exception: genuinely free and unlimited, with the cost shifted to your setup time, your electricity and your patience.

Fitting transcripts into a real workflow

A transcript nobody reads is wasted effort. The value appears when the text goes somewhere useful automatically.

For remote teams

Send the summary and action items into your project tool or team chat right after the call. Most transcription tools offer native integrations or a Zapier/Make connection, and the same no-code thinking applies to the rest of your admin — see this guide on automating repetitive business tasks without coding. A simple rule — “transcript summary posts to the project channel, full transcript lives in the shared drive” — means people who missed the call can catch up in two minutes.

For solopreneurs

Client calls become proposal drafts, scope documents and follow-up emails. Paste the transcript into an AI chat assistant and ask for a bullet summary of commitments made, then check it against the transcript yourself before sending anything.

For content creators

Internal discussions are raw material. A recorded strategy session can become a blog outline, a newsletter section or subtitle files for a clip.

Always add a review pass

Skim the transcript against the timestamps for any decision, number or commitment that matters. AI misheard words look perfectly confident on the page, which is exactly what makes them dangerous.

Frequently asked questions

How accurate is AI meeting transcription?

Accuracy varies a lot with audio quality, accents, crosstalk and specialist vocabulary. Clear single-speaker audio with a headset performs best; a noisy room with several people talking over each other performs worst. Always review before relying on a transcript.

Can I transcribe a meeting for free?

Yes. Self-hosted Whisper is free and has no minute limits, and most commercial tools offer a free tier with capped monthly minutes, limited uploads and restricted exports. Read the specific limits before planning a workflow around a free plan.

Do I need permission to record and transcribe a meeting?

Recording consent rules differ by country and sometimes by region within a country. Announce recording at the start of every call and check the rules that apply to you and your participants rather than assuming.

Will transcription work for languages other than English?

Most major tools support many languages, though quality is usually strongest in English and other widely represented languages. Test a short sample in your language before committing to a paid plan.

Related guides

• How to Transcribe Videos with AI: A Complete Beginner’s Guide to Auto Subtitles
• Complete Guide to AI Voice Tools: Text to Speech, Voice Cloning and Speech Recognition for Beginners
• Complete Cloud Storage Setup Guide: Automate Backups and Secure Your Files