Best AI note takers with speaker identification in 2026
Speaker labels split a transcript into Speaker 1 and Speaker 2. Speaker identification puts real names on those lines. On a video call the names come from the meeting app; in a room there is no participant list, so the tool has to recognize voices. Wave AI labels speakers in 76 languages, and its Voice ID can learn and name the voices of people you meet often. (We make Wave.) Otter.ai also learns voices you tag, in six languages.
Prices checked on November 18, 2026, on each tool's own website.
Compared side by side
| Tool | In a room | On video calls | Learns voices across meetings | Languages | Cheapest paid plan |
|---|---|---|---|---|---|
| Wave AI | Speaker labels from a phone or watch | Speaker labels, with a bot or no bot | Yes, Voice ID names people you meet often | 76 | Pro: $80 a year on the web |
| Otter.ai | Speaker identification in its apps | Speaker identification | Yes, after you tag a speaker | 6 | Pro: $8.33 a month, billed yearly |
| Fireflies.ai | Speaker 1, Speaker 2 | Names, when its bot or Chrome extension is used | Not listed by Fireflies | 100+, most in beta | Pro: $10 a month, billed yearly |
| Fathom | iPhone app; names only you | Every line attributed to a speaker | Only your own voice | 7 | Premium: $16 a month, billed yearly |
| Notta | Speaker 1, Speaker 2 | Speaker 1, Speaker 2 | Not listed by Notta | 58 | Pro: $97.99 a year |
| Plaud | Speaker 1, Speaker 2 | Speaker 1, Speaker 2 | Only your own voice | 112 | Device from $159, plus a plan |
Wave AI
Best for: in-person meetings with people you meet again and again.
Wave records in-person meetings from an iPhone or Android phone, an Apple Watch or a Wear OS watch, with no bot and no extra hardware, and returns a speaker-labeled transcript in 76 languages. Voice ID can learn the voices of people you meet often and name them in later recordings. Wave keeps the original audio next to the transcript, so you can play back a line to check who said it. The same account records phone calls on both sides and Zoom, Google Meet and Microsoft Teams.
Limits: Voice ID has to learn a voice before it can name it, so a new person appears as a numbered speaker first. Wave does not show a live transcript during the meeting. The free plan is 30 minutes a month.
Price: Pro is $15 a month or $80 a year on the web, with unlimited recording. See Wave pricing.
Otter.ai
Best for: live transcripts with speaker names, in English.
Otter includes speaker identification on every plan, including the free one, and shows a live transcript during the call (Otter pricing). Once you tag a speaker, Otter builds a speaker profile and labels that person automatically in later recordings (Otter help). If you need a live transcript, Otter is the better choice over Wave.
Limits: Otter transcribes six languages. Free conversations stop at 30 minutes and Pro conversations at 90.
Price: Pro is $16.99 a month, or $8.33 a month billed yearly.
Fireflies.ai
Best for: names on video calls across many meeting apps.
Fireflies sends its bot into Zoom, Google Meet, Microsoft Teams and Webex and transcribes in 100+ languages, most of them in beta (Fireflies pricing). Speaker name detection works when its bot joins the call or you use its Chrome extension (Fireflies guide).
Limits: outside those two cases you get generic labels. Uploaded files use Speaker 1 and Speaker 2 until you rename them by hand, so a recording made in a room arrives without names.
Price: Pro is $18 a seat a month, or $10 a month billed yearly.
Fathom
Best for: free video calls with every line attributed.
Fathom's free plan records and transcribes Zoom, Google Meet and Microsoft Teams calls with no limit (Fathom pricing). Fathom says it attributes every line to the right speaker, with a bot or bot-free (Fathom transcription). Its iPhone app records in-person meetings on every plan.
Limits: in person, Fathom creates a voiceprint only of you, if you opt in. Other people in the room may appear unnamed in the transcript, and the app is not available on Android (Fathom help). Fathom transcribes seven languages.
Price: Premium is $20 a month, or $16 a month billed yearly.
Notta
Best for: speaker-labeled transcripts in many languages on a low-cost plan.
Notta lists speaker identification on every plan, including Free, and transcribes 58 languages in real time (Notta in-person). Pro is $97.99 a year with 1,800 minutes a month (Notta pricing).
Limits: Notta labels speakers as Speaker 1, Speaker 2 and so on, and you rename them yourself (Notta support). The free plan stops each conversation at 3 minutes.
Price: Pro is $97.99 a year. Business is $199.99 a year.
Plaud
Best for: people who want a dedicated recorder instead of a phone app.
Plaud sells recorders from $159 to $189: cards that attach to the back of a phone and wearable pins (Plaud devices), and transcribes 112 languages. Its free Starter plan includes 300 minutes of transcription a month, and Pro is $8.33 a month billed yearly for 1,200 minutes (Plaud plans).
Limits: Plaud labels people as Speaker 1 and Speaker 2. It recognizes only your own voice once you add a voice profile, and you rename everyone else by hand (Plaud support). You also have to buy and carry the device.
Price: $159 to $189 for the device, then the Starter, Pro or Unlimited plan.
What affects accuracy
- Distance from the microphone. Voices far from the phone or recorder are quieter and harder to tell apart. Put it in the middle of the table.
- Crosstalk. When two people talk at once, any tool can merge them into one speaker.
- Number of speakers. More people means more speaker changes and more chances to mix up similar voices.
We have not run accuracy tests, so this page does not rank the tools on accuracy. Check names against the audio before you share a transcript.
How to choose
- In-person meetings with the same people each week: Wave, for Voice ID across recordings.
- Video calls in English with a live transcript: Otter.
- Free video calls where the meeting app supplies the names: Fathom.
- Many languages on a budget: Notta, if you are happy to rename speakers yourself.
- A separate device: Plaud.
Voice recognition handles people's voices, so tell everyone before you record. Consent laws differ by place.
