Otter.ai accuracy: what Otter says affects it, and the settings that improve it

By · Published

Otter publishes no accuracy percentage in its help center. It says its speech recognition is generally quite accurate, more so with Otter Notetaker, which records a meeting's own audio stream, than with manual recording through a microphone, and that results vary with background noise, accents and the complexity of the conversation. Its advice comes down to a cleaner recording, tagged speakers, and custom vocabulary for names and jargon.

I built YapToText, a dictation app for the Mac, so read what I say about Otter with that in mind. Every fact about Otter here comes from its own help center, read on October 11, 2026.

I publish no accuracy numbers either, for any app.

Download YapToText on the Mac App StoreThis app is completely free. You don't have to make an account. There are no subscriptions, no difficult downloads, and nothing is locked. Made for Apple silicon Macs. Source on GitHub.

Make the recording better

  • Cut background noise. Otter calls it generally the biggest factor in inaccurate or missing transcription.
  • Avoid talking over each other. Overlapping speech may be transcribed wrongly, or not at all.
  • Use an external microphone. Omnidirectional for several people in a room, unidirectional for your own voice.
  • Place it close. Within three feet of the person speaking, and do not move or touch it.
  • Speak normally. Clearly, at your usual pace.

Teach Otter your words

Custom vocabulary is for proper names, jargon, acronyms and unusual spellings. How much you can add depends on the plan.

From Otter's Manage vocabulary article, read October 11, 2026.

PlanCustom vocabulary
Basic, freeUp to 5 words or phrases
ProUp to 100 names and 100 other terms
BusinessA team vocabulary of 800 names and 800 other terms, plus 200 and 200 per user
  1. Click your name in the top left and open Account settings, General.
  2. Click Manage Vocabulary.
  3. Choose Names of people, or Other vocabulary, and add the term.

Tag speakers

Tagging trains Otter to recognize voices, which improves who-said-what in later conversations. The speaker names page has the steps.

How it works, in Otter's words

  • Cloud only. Otter says it is entirely cloud-based and uses no human transcribers.
  • Filler words. Sounds such as um, ah and hmm are ignored automatically, even if you add them to your vocabulary.
  • Review it. Otter says it is good practice to review and edit transcripts, especially for important conversations.
  • Languages. Six. The languages page lists them.

Why Notetaker beats a microphone

A microphone in a room hears the room. Notetaker and the desktop app take the meeting's audio directly, which removes the speakers, the echo and most of the noise from the path. If accuracy is poor when you record by hand in a browser, change the method before anything else. The page on Otter not recording covers the browser limits.

Accuracy for your own dictation

Otter is built for conversations between several people. For your own voice into a document, the same principles apply and are easier to meet: one speaker, one microphone, close. YapToText, the free Mac app I built, adds dictionaries that prime its speech model with your names and terms, and runs on the Mac. Improving dictation accuracy on a Mac and choosing a microphone for dictation go through it.

Download YapToText on the Mac App StoreThis app is completely free. You don't have to make an account. There are no subscriptions, no difficult downloads, and nothing is locked. Made for Apple silicon Macs. Source on GitHub.

The short version

  • Biggest factor. Background noise.
  • Microphone. External, within three feet.
  • Vocabulary. 5 terms on Basic, 100 names and 100 terms on Pro.
  • Method. Notetaker or the desktop app, over a browser recording.
  • Always. Review the transcript.

If Otter changes its guidance, open an issue on GitHub and I will update this page. Every one gets read.

Download YapToText on the Mac App StoreThis app is completely free. You don't have to make an account. There are no subscriptions, no difficult downloads, and nothing is locked. Made for Apple silicon Macs. Source on GitHub.

About the author. Ryleigh Newman is a developer whose hands make typing difficult and who dictates every day with YapToText, the app this site describes. Facts about other apps come from their own sites, with the date they were read. More about the project.

Sources

Questions

How accurate is Otter.ai?

Otter gives no percentage in its help center. It says accuracy is generally good, better with Notetaker than manual recording, and varies with noise, accents and the conversation.

How can I improve Otter's accuracy?

Reduce background noise, avoid overlapping speech, use an external microphone within three feet, tag speakers and add custom vocabulary.

How many custom words can I add to Otter?

Up to 5 on the free Basic plan, and up to 100 names and 100 other terms on Pro.

Does Otter remove filler words?

Yes. Otter says sounds such as um, ah and hmm are ignored automatically.

Does Otter use human transcribers?

No. Otter says it is entirely cloud-based and uses no human transcribers.