Echoon User Guide

Echoon turns speech into polished, useful text — live from a microphone or meeting, or from a recording — with AI corrections, translation, summaries, and assistants on top. This guide covers what every mode shares first, then what makes each mode special.

Getting started

Sign in, or try as a guest

You can look around anonymously, but transcribing requires an account: guest transcripts live only in the current browser session and guest accounts get neither free transcription nor free credits. Sign in — Microsoft (Azure AD work/school, recommended for companies), Google, Apple, GitHub, Facebook, X, or LINE — to transcribe free, receive your weekly free credits, keep your history, and sync across devices.

Credits & pricing

  • Live transcription is free on one audio source, forever, with a signed-in account. Credits pay for AI: 1 credit = 1 hour of it, and lists at US$0.30 — one flat rate, no subscription.
  • Advanced and Live Captions run at 2x, Meeting Bot at 3x (the notetaker bot costs extra), Voice Memo and File Transcript at 1x — which already includes one AI pass. In every live mode the first audio source is free and each source past it adds 1x, so free transcription of your mic and your speaker at the same time costs 1x.
  • AI-enhancing a finished transcript costs 1x of its audio length — free, though, on a Voice Memo or File Transcript, where it is already part of the 1x.
  • Rates are quoted per hour, but you are charged by the second on the audio you actually record or upload: stop after 90 seconds and you pay for 90 seconds. There is no minimum charge and no rounding up to the next minute (AI enhancement is the one exception — it bills a minimum of one minute).
  • We run bonus-credit campaigns rather than discounts: while one is on, the bigger packs grant extra credits on top of what you paid for. The purchase screen names the campaign, shows the bonus on each pack, and says when it ends. Bonus credits are ordinary credits — same value, same uses, no expiry.
  • There are three kinds of credits, spent in this order. Trial credits are the weekly free allowance: they reset at the start of your local week, and one session can use at most 0.5 of them (15 minutes of Advanced) with a short wait before the next — a session that hits that limit keeps running on your other credits, and plain transcription keeps running either way because it is free. Reward credits (welcome grant, first-purchase bonus, referrals) never expire and have no session limits. Purchased credits have no limits at all, and are the only ones that can run the Meeting Bot, which uses an outside provider we pay by the hour.
  • Your balance is shown in the bottom-left menu — tap it to see the ledger or buy packs.

Refunds

Credits are delivered the moment payment clears, so purchases are non-refundable as a rule — but a purchase whose credits you have not used at all is refunded in full within 14 days. Once any of them have been used, that purchase is no longer refundable; there is no partial refund. Only the purchased credits count: spending your weekly trial allowance or a reward after buying does not affect your refund, because those are used up first. The full rules are on the Refund Policy page, linked from the account menu.

Your first transcript

Pick a mode from the Home screen or the sidebar, choose the recognition language, and press start (or drop a file). Every session is saved to History in the sidebar, where you can search, rename, share, or delete it.

Session settings — shared by every mode

The setup panel in front of each live mode (and the settings drawer during a session) is built from the same blocks. A minute spent here pays off directly in accuracy.

Recognition language

Echoon recognizes 29 languages. Set your default in Settings → Speech recognition; each session can override it, and Bilingual Talk detects the language of each sentence automatically.

Audio sources: microphone & speaker

Live modes can record your microphone, the speaker/system audio, or both — so an online call is captured from both sides, not just yours. Speaker capture opens a screen-share prompt: pick the tab or screen playing the audio and tick "Share audio".

Scene × industry

Tell the AI where the conversation happens (one scene) and which industries it touches (multi-select — an IT project for a hospital can be both). Curated term lists load, and corrections, translation wording, and note-taking all adapt. Both default to Auto: leave them there and the AI names the setting from the conversation itself once it has heard enough, then works from it for the rest of the session. Add your own scenes and industries in Settings; any language works.

Hotwords & personal dictionary

Names, products, and jargon come out spelled right when recognition is biased toward them. Use the per-session hotwords box for one-off terms, and Settings → Dictionary for the terms you always need.

Background documents

Paste context or upload a PDF, Word, or text file — an agenda, resume, or product doc. Text is extracted on your device, long documents are AI-compressed into a digest, and everything grounds the AI: corrections, Q&A, insights, and speaking suggestions all get sharper.

Speaker labels & voice enrollment

Turn on speaker labels to separate who said what. Enroll a voice once in Settings → Speech recognition (read the sample passage for 10–30 seconds) and transcripts show the real name instead of "Speaker 1" — you can also enroll a speaker straight from a finished transcript. Extra samples per person make matching noticeably more reliable.

The AI assistant

In Advanced and Meeting sessions the assistant works in real time. Standard, Voice Memo, and File transcripts get the same panel after a one-tap AI enhancement.

Summary

A structured key-point digest — decisions, action items, open questions, and more, with sections that adapt to your scene — builds while people are still talking and is finalized the moment the session stops.

Q&A — ask the conversation

Ask anything about what has been said so far. With question auto-detection on, Echoon notices when the other side asks you something and drafts an answer automatically; every exchange is saved to the Q&A tab.

"What to say?" suggestions

Stuck for words mid-conversation? One tap gets reply suggestions grounded in your background documents and the conversation so far.

Insights

One tap analyzes the conversation to date — open questions, risks, and action items, tailored to the scene and your background material.

AI enhance (1x)

Any finished Standard, Voice Memo, or File transcript can be upgraded afterwards with sentence corrections, translation, and a summary. Flip on auto-enhance in the setup panel to run it automatically when the session ends. If enhancement fails, nothing is charged.

The modes, one by one

Six ways in, all built on the same foundation — pick by how the audio arrives and how much AI you want on top.

Standard Transcript Free

Plain realtime transcription of your microphone and/or speaker audio — the fastest, cheapest way to get accurate words on the page.

  • Best for interviews, dictation, and any time you just need the text.
  • Hotwords, scene/industry, and speaker labels all work here too — only the AI copilot is off.
  • You can AI-enhance the finished transcript later for 1x, so starting in Standard never locks you out of summaries.

Advanced Transcript 2x

Realtime transcription plus the full AI copilot: sentence-by-sentence corrections, live translation, a running key-point digest, and Q&A, insights, and speaking suggestions on demand.

  • Set the scene, industries, and background documents before starting — every AI feature gets sharper.
  • The key-point digest is ready the moment the meeting ends; no waiting for minutes.
  • Turn on question auto-detection when you expect to be asked things — answers are drafted for you as questions land.

Bilingual Talk 2x

A realtime two-way conversation between two languages — put the screen between you and whoever you do not share a language with.

  • Pick the language pair up front. Each sentence is language-detected on its own and translated into the other half of the pair, so neither side has to switch anything while talking.
  • A third language nobody selected still works — it is translated into language B rather than dropped.
  • Adjust the font size and go fullscreen from the conversation view; a single line of background ("buying a camera at an electronics store in Osaka") noticeably improves translation.

Meeting Bot 3x

Paste a Zoom, Google Meet, or Teams link and an AI notetaker bot joins the call, relaying the meeting audio into the Advanced pipeline — corrections, translation, digest, and all assistants included.

  • Ask the host to admit the bot when it knocks; you can change its display name before it joins.
  • Keep the Echoon tab open for the whole meeting — the bot relays audio, and transcription runs in your browser.
  • The extra credit (3x vs Advanced’s 2x) covers the notetaker bot itself.

Voice Memo 1x

Record locally in the browser, then transcribe in one batch pass when you stop.

  • Nothing is uploaded until you press transcribe; you can pause and resume, download the audio, or discard and re-record freely.
  • Closed the tab by accident? Untranscribed recordings are offered for recovery the next time you open Voice Memo.
  • Auto-enhance gives you corrections and a summary the moment transcription completes — already included in the 1x.

File Transcript 1x

Upload an existing audio file and get the full transcript back at the Standard rate.

  • MP3, WAV, M4A, WebM, FLAC, and OGG are supported.
  • Pick the recognition language before uploading for the best accuracy.
  • Auto-enhance adds corrections, translation, and a summary as soon as the transcript lands — already included in the 1x.