Krisp Accent Conversion: Daily Hours, Plans and Setup Guide

Krisp's accent conversion reshapes spoken English in real time on your Mac or Windows computer, and it works in two directions. The speaker side changes how your own accent reaches everyone on the call. The listener side, which Krisp also calls Accent Understanding, adapts other people's accents for your ears only. In the Meeting AI plans the output is always US English. Core gives you 1 hour a day in total, which Krisp's help says is shared between both sides, while Advanced gives 4 hours a day on the speaker side and unlimited listening ^[1]. Core costs $8 per user per month billed annually, or $16 per user per month billed monthly. Setup takes a script reading, a decent headset and two device choices. It never works offline.

Your accent going out, their accents coming in

The two halves ride on the two virtual devices Krisp installs. Speaker-side conversion sits on Krisp Microphone, so the calling app receives your converted voice and every participant hears it. Listener-side conversion sits on Krisp Speaker and changes nothing for anyone else: only you hear the adapted audio, and it applies to every participant who talks to you. Krisp describes the result as refining your speech while keeping your natural voice. During a call you control each half separately, with the Convert My Accent and Convert Others Accent toggles.

You do not pick an accent first. Krisp's Meeting AI help says a single Universal Accent model converts any accent to US English ^[2]. Results still vary by origin. For your own voice, the Core pricing card names Indian, Filipino, South African and Latin American accents, and Krisp's speaker-side page lists LatAm, Indian and Filipino English, available from app version 2.57.8. For other voices, the listener page reports its strongest results with Indian, Filipino, Latin American, African and Mandarin Chinese accents. Krisp lists British English output only as a Call Center AI option.

Processing stays on your machine. The listener page says the adaptation runs on the computer's CPU with roughly 200 milliseconds of delay or less, and that the audio is neither stored nor sent to outside servers. That local design has one limit. Krisp's speaker-side page says accent conversion does not work offline, and Krisp Microphone and Krisp Speaker appear in other apps only while you are signed in to the desktop app. Plan on a connected, signed-in computer for every call where you need conversion.

How many converted minutes each plan allows

Core's single hour is a pool, per Krisp's help: minutes on either side draw from the same daily allowance. Consider a typical day. A 45-minute client call plus a 30-minute standup comes to 75 minutes, already past the cap if conversion stays on throughout. By our arithmetic, Core covers 5 hours per user across a five-day work week, while Advanced covers 20 hours per user of your own converted speech over the same week, with listening uncounted. Krisp does not publish what happens when the daily allowance runs out, so plan as if conversion stops there.

The step from Core to Advanced costs $7 per user per month billed annually, which is $84 per user a year; on monthly billing the gap is $14 per user per month, or $168 per user a year. That buys three more speaker-side hours a day and unlimited listening, alongside Salesforce, manager view and the team admin controls. If your converted minutes on both sides together regularly run past an hour a day, Core cannot cover them. Advanced fits heavy speakers, and anyone who wants other people's accents converted on every call. Below an hour a day, the extra hours sit unused.

The billing term matters more than the plan if you only need conversion for a few months. By our arithmetic, monthly billing beats a year billed annually only if you stop within five months: five months of Core billed monthly cost $80 per user, against $96 per user for a year billed annually. At six months, Core billed monthly reaches $96 per user, the same as a year billed annually, by our arithmetic. Our Krisp pricing guide works through that rule and the refund terms.

Hardware and audio routing come first

Accent conversion runs in real time, and Krisp warns that a computer below its minimum requirements can lose audio quality without ever showing high CPU use ^[3]. Krisp's recommended desktop baseline is a Mac with an Intel Core i5 or Apple silicon and 8 GB of RAM on macOS 14 Sonoma or later, or a 64-bit Windows 10 or 11 PC with an Intel Core i5 or AMD Ryzen 5 and 8 GB of RAM. Update the app to at least version 2.57.8 before relying on the speaker side.

Headset choice is the next lever. Krisp recommends a wired USB headset with a boom microphone, because a wireless connection adds latency to speech that is already being rebuilt as you talk. Bluetooth headsets still work. Expect some added delay, though, and test the converted sample on the headset you will actually wear. No expensive hardware is required, Krisp's audio guidance adds; the boom and the cable matter more than the price tag.

Then route the calling app through Krisp. In Zoom, Microsoft Teams, Slack or a desktop softphone, choose Krisp Microphone as the input and Krisp Speaker as the output: the speaker side needs the first, the listener side the second. Turn off the calling app's own noise suppression and echo cancellation so Krisp is the only processor on your voice; in Zoom, Krisp's help says to set background noise removal to Low instead. Browsers are the weak spot. Chrome, Edge and Opera let a web app use Krisp Microphone, but Krisp Speaker cannot be used in Firefox or Safari, so listener-side conversion there is out of reach. Our noise cancellation setup guide covers each app's settings.

Record your sample, then switch it on mid-call

Speaker-side setup starts with your own voice. Krisp's accent page asks you to read a provided script into the microphone, and the Test Accent Conversion sample you record is processed locally and never sent to or stored on any server. Listen back before a real meeting, with the same headset and in the same room you will call from. If the converted sample sounds clipped or uneven, fix the hardware and gain settings described below before judging the model itself. The full sequence takes four steps:

1. Update Krisp, sign in to the desktop app and select your physical headset inside it.
2. Choose Krisp Microphone and Krisp Speaker in the calling app.
3. Record the script sample and listen to the converted version.
4. In the call, turn on Convert My Accent, Convert Others Accent, or both.

Choosing which toggles to use is where Core users save minutes. Both halves draw from the same daily hour, so in a meeting where you mostly listen, leaving your own side off stretches the allowance. Switch on only what the call needs. Advanced changes the calculus, because listening is unlimited there: leave Convert Others Accent on for the whole day and budget the four speaker-side hours for the calls where you do most of the talking, such as demos, interviews or client reviews.

Habits that keep converted speech natural

Automatic gain control causes most quiet failures. Browsers and softphones that keep pushing your microphone level up and down can hurt accent conversion, Krisp's help warns, and the Lock at optimal level setting for Krisp Microphone volume exists to stop that, though only on Windows. Mac users should switch off gain control inside the calling app instead; Slack, for example, has its own automatic setting for calls and huddles. Krisp recommends a USB wired headset with a boom mic for accent conversion.

Streaming and recording follow the same routing rule. Krisp's help says conversion applies only when Krisp Microphone is the selected input, so a streaming or recording app pointed at your physical microphone captures your original accent. Check the input in every such tool once, not just in your meeting apps. On a laptop, close heavy apps before long calls. Because the model degrades quietly on an overloaded machine instead of warning you, we suggest listening to the Test Accent Conversion sample again after any new headset, laptop or operating system update.

YouTube, live translation and other look-alikes

Krisp says listener-side accent conversion on YouTube videos is free (November 2026), and its Chrome Web Store listing says it needs no subscription or credit card ^[4]. It changes only what you hear, with US English as the output, and does nothing for your voice in meetings. It is not a plan. Conversion on calls runs on Core or Advanced after the 7-day free trial, which needs no credit card either.

Searches for Krisp voice translation or live translation land on a different product. Real-time speech-to-speech translation, formerly called AI Live Interpreter, is an add-on to Krisp's Call Center AI, covering more than 60 languages and arranged through an account executive or Krisp's sales team. Developers reach the same capability as a Voice Translation API in 61 languages. Core, Advanced and Enterprise list no translation, and unlike accent conversion it runs in Krisp's cloud. Call Center AI is priced per agent and sold separately, as our call center guide explains. Discord's built-in Krisp, finally, has no accent conversion at all.

The bottom line

Krisp's accent conversion is worth a trial if accents slow down your calls in either direction, and the plan choice comes down to minutes on both sides together. Core's shared hour suits someone who converts under an hour a day, for example one short daily call; anyone speaking through it for most of the workday, or wanting others' accents converted on every call, needs Advanced at $15 per user per month billed annually. Use the 7-day free trial, which asks for no credit card, to record your sample on your real headset. Krisp's trial card lists accent conversion as limited, so judge quality there rather than timing a full day. Skip it if you need offline use or live translation inside meetings.