Meeting transcription with speaker separation, from a recording you already have

sAInote turns a meeting recording you already made into a readable transcript with the speakers separated, and then into structured minutes rather than one unbroken block of text. Nothing joins your call: no bot in the participant list, no calendar to connect, no access to your meeting account. You record with your phone, your laptop, a room mic, or your conferencing tool's own local recording, then upload the file. Recordings up to 20 minutes are processed immediately on every plan; recordings from 20 minutes to 2 hours run in the background on Pro.

Transcribe a meeting 15 free minutes, one time · no account needed

What you get from a meeting recording

Speaker separation on long recordings: Long meetings are split by speaker, so an hour of back-and-forth reads as distinct turns instead of one paragraph. sAInote distinguishes voices rather than identifying people by name, so you decide the labels.
Detail that survives to the end: Long recordings are summarised hierarchically: the audio is processed in chunks and then condensed in layers, so a decision made in the last ten minutes of a two-hour meeting is kept as faithfully as one made in the first.
Up to 2 hours, processed in the background: Recordings between 20 minutes and 2 hours are a Pro feature and run in the background. Start the upload, close the app, come back to a finished note. Two hours is the hard cap for a single recording.
Your minutes format, not a generic summary: With Pro Assistant you write and save your own instruction — decisions, owners, action items, risks, in the order your team expects — and apply it to any recording instead of accepting one fixed summary shape.
Attendee and product names spelled right: Special Names is a personal spelling glossary on Pro. Add colleagues, clients, products and internal jargon, and those words are spelled correctly instead of guessed phonetically. It is a spelling reference only and never rewrites what was said.
Minutes in another language: Edit & Translate handles editing and translation in one pass into roughly 35 target languages, in Smart Summary mode. The language you pick is remembered, and translated notes carry a badge in your history.

How to transcribe a meeting recording with speaker names

1

Record or export the meeting

Use your phone on the table, your laptop, a room mic, or your conferencing tool's own local recording. Export as WAV, MP3, M4A, MP4, WebM, OGG or Opus, up to 64 MB.
2

Upload the file

Open sainote.com in a browser, or use the iPhone, iPad or Mac app. On Mac you can also capture a quick note with the global shortcut Control+Option+Command+N without opening a window.
3

Pick a processing mode

Choose Quick Note for a verbatim transcript, Smart Summary for an edited version with headings and lists, or Pro Assistant to apply your own saved minutes instruction.
4

Let long recordings finish in the background

Anything under 20 minutes comes back straight away. Between 20 minutes and 2 hours, Pro processes the recording in the background, so you can close the app and check later.
5

Find it again later

Signed-in users get searchable history across devices. Search matches both the raw transcript and the processed minutes, so a half-remembered phrase from a meeting is enough to find the note.

Meeting transcription with speaker separation, without a bot on the call

sAInote, an AI voice-to-text app, works from a file. It does not join meetings. No assistant appears in the participant list, no calendar has to be connected, and no access to your meeting account is requested. For a lot of teams that is the deciding factor: the recording is something you already control, and the people in the room see nothing new on screen.

In practice this means the source can be anything. A phone left face-up on a conference table, a laptop mic in a one-to-one, a dedicated room recorder, or the local recording your conferencing tool already saves to disk when you press record. Export that file and upload it in the browser, or from the iPhone, iPad or Mac app.

The honest trade-off is that there is no live transcript scrolling while people are still talking. You get the transcript after the recording is uploaded, not during the meeting. The upside is that in-person meetings, workshops, site visits and interviews are covered exactly as well as calls are, because sAInote never needed to be inside the call in the first place.

How a two-hour meeting is handled

Two things happen to a long recording. First, the speakers are separated, so the transcript reads as turns rather than a continuous wall of text — you can see where one person stops and another starts, which is what makes an hour of discussion skimmable at all. sAInote separates voices; it does not know anyone's real name from the audio, so the labels are yours to assign.

Second, long recordings are summarised hierarchically. The audio is processed in chunks and those chunks are condensed in layers, rather than everything being squeezed into one pass. That is the difference between minutes where the last twenty minutes collapse into a single line and minutes where a decision made at the end carries the same weight as one made at the start.

The limits are worth stating plainly. Recordings up to 20 minutes are processed immediately on every plan, including guests. Recordings from 20 minutes to 2 hours are Pro only and are processed in the background. Two hours is the hard cap for a single recording, and anything longer is rejected. The upload cap is 64 MB, which usually matters more than length for a long meeting: export compressed voice-quality audio rather than uncompressed.

Minutes in your format, with names spelled correctly

A generic summary is rarely what a meeting needs. Most teams have a shape they expect: decisions at the top, owners next to action items, open questions at the bottom, and everything else compressed. Pro Assistant lets you write that instruction once, save it, and apply it to any recording. The output follows your template instead of a fixed house style.

Smart Summary is the middle ground and is available on every plan, including guests. It removes filler, repairs half-finished sentences, and adds paragraphs, headings and lists, without changing what was actually said. Quick Note is the other end: a verbatim transcript with nothing added and nothing removed, which is what you want when the exact wording of a commitment matters.

Special Names covers the part transcription usually gets wrong. It is a personal spelling glossary on Pro: add the names of colleagues, clients, products and internal jargon, and those words are spelled the way you spell them instead of being guessed phonetically. It is a spelling reference only — it never changes the meaning of what was said, and your glossary is never shown anywhere.

Plans, languages and what happens to your audio

Usage is measured in minutes of audio, not per request or per word, and the mode you choose does not change the cost: a 3-minute recording costs 3 minutes of balance whether it comes back as a verbatim transcript or as formatted minutes. Guests get 15 minutes in the browser with no account, once — not monthly. A free account also gets 15 minutes, once, plus searchable history across devices. Pro renews 360 minutes, six hours, every month, and adds Pro Assistant, Special Names, recordings over 20 minutes and priority support.

Signing in has no password. You get a 6-digit code or a magic link by email, or you sign in with Apple or Google. The interface ships in 11 languages — Arabic, Bengali, English, French, Hindi, Portuguese, Russian, Simplified Chinese, Spanish, Turkish and Urdu, with Arabic and Urdu rendered right-to-left. Speech recognition covers considerably more languages than the interface does.

Meeting audio is kept in private cloud storage. It is never public, never indexed and never shareable by URL. Recordings are retained for processing and cleaned up on a schedule, and accounts can be deleted in the app, which anonymises the data. For teams building on the REST API, neither audio nor text is stored for API traffic at all.

Meeting transcription questions

Can sAInote separate speakers in a meeting recording?

Yes, speakers are separated on long recordings, so an hour of discussion reads as distinct turns rather than one paragraph. sAInote distinguishes voices rather than identifying people by name, so you assign the labels yourself. Separation happens on the file you upload, whether that comes from the browser or the iPhone, iPad or Mac app.

Does a bot join my video call to record it?

No. sAInote never joins a call, never appears in the participant list, and does not connect to your calendar or your meeting account. You make the recording yourself — a phone on the table, a laptop, a room mic, or your conferencing tool's own local recording — and upload the file afterwards.

Can I transcribe a 2 hour meeting?

Yes, 2 hours is the maximum length for a single recording, and anything longer is rejected. Recordings up to 20 minutes are processed immediately on every plan. Recordings between 20 minutes and 2 hours are a Pro feature and are processed in the background, so you can close the app and come back to a finished note.

Does sAInote transcribe live while the meeting is happening?

No. sAInote transcribes a recording, not a live conversation, so there is no real-time transcript appearing while people talk. Record with whatever you already use, then upload the file. The trade-off buys you coverage of in-person meetings, workshops and interviews, not only calls made through a conferencing tool.

What file formats and file size can I upload?

WAV, MP3, M4A, MP4, WebM, OGG and Opus, up to 64 MB per file. For long meetings the size cap usually bites before the length cap: a two-hour recording still has to fit inside 64 MB, so export it as compressed voice-quality audio rather than uncompressed.

How do I get decisions and action items instead of a generic summary?

Write your own instruction with Pro Assistant and save it. Describe the minutes format your team expects — decisions first, owners beside action items, open questions at the end — and apply that saved instruction to any recording. Pro Assistant is a Pro feature; Quick Note and Smart Summary are available on every plan.

Will attendee and client names be spelled correctly?

Add them to Special Names, a personal spelling glossary available on Pro. Names of colleagues, clients, products and internal jargon are used as a spelling reference, so they appear the way you write them instead of being guessed phonetically. It affects spelling only and never rewrites what was said.

Is there a free way to try meeting transcription?

Yes, 15 minutes of audio, one time. Guests get 15 minutes in the browser with no account and no history. A free account also gets 15 minutes once — not monthly — plus searchable history across devices. Pro renews 360 minutes, six hours, every month and adds recordings longer than 20 minutes.

Can I get the minutes in a different language than the meeting?

Yes, Edit & Translate edits and translates in a single pass into roughly 35 target languages, and it works in Smart Summary mode. Your chosen language is remembered, and translated notes are marked with a badge in your history. The interface itself ships in 11 languages; speech recognition covers considerably more.

Start with the recording you already have

Upload a meeting file in the browser or from the iPhone, iPad or Mac app, and see the transcript and minutes it produces. Fifteen minutes of audio, one time, without an account.

Transcribe a meeting