sAInote

How to transcribe a long audio file, up to 2 hours

sAInote transcribes a single audio file of up to 2 hours; anything longer is rejected and has to be split before it will run. Recordings up to 20 minutes are processed straight away on every plan, including guests without an account. Anything from 20 minutes to 2 hours is processed in the background and needs Pro; you can close the app and come back to the finished note. Every file has to be 64 MB or smaller, whatever its duration. sAInote is an AI voice-to-text app for the web, iPhone, iPad and Mac.

Start a long recording Without an account you get 15 minutes of audio, one time. From 20 minutes to 2 hours needs Pro.

What happens to a long recording

Under 20 minutes: processed immediately
Recordings up to 20 minutes are processed as soon as you stop recording or finish uploading, on every plan, including guests with no account. Quick Note and Smart Summary are both available at that length.
20 minutes to 2 hours: background, Pro only
Longer recordings are queued and processed in the background on a Pro account. You can close the browser tab or quit the app; the finished note appears in your searchable history when it is ready.
Built up in pieces, not one pass
A two-hour recording is not pushed through a single pass. It is split into sections, each section is worked through, and the results are built up, so the last twenty minutes get the same attention as the first twenty.
Speaker separation on long audio
Long recordings get speaker separation, so a two-hour interview or meeting reads as distinct voices instead of one undivided block of text that you have to untangle line by line afterwards.
Special Names for recurring names
Special Names is a personal spelling glossary on Pro, and a spelling reference only. It matters more on long audio, where one mis-spelled surname or product name can recur dozens of times in a single note.
64 MB per file, whatever the length
The upload limit is 64 MB regardless of duration. Accepted formats are WAV, MP3, M4A, MP4, WebM, OGG and Opus. At two hours, a compressed format is what keeps a file under the cap.

How to transcribe a 1 hour or 2 hour recording

  1. 1

    Check the length and the file size first

    Confirm the recording is under 2 hours and the file is 64 MB or smaller. Anything longer than 2 hours is rejected, so split it into parts before you upload.

  2. 2

    Sign in to a Pro account

    Sign in without a password using a 6-digit emailed code, a magic link, or Apple or Google. Anything over 20 minutes needs Pro, so a guest session will not carry a 1-hour file.

  3. 3

    Choose your processing mode

    Pick Quick Note for a verbatim transcript, Smart Summary for an AI-edited version with paragraphs, headings and lists, or a Pro Assistant instruction you have written and saved yourself.

  4. 4

    Upload or record, then walk away

    Once the upload has finished, or you have stopped the recording, you can leave it to process. Long files are processed in the background, so closing the browser tab, the iPhone app or the Mac app does not cancel the job.

  5. 5

    Open the finished note from your history

    Come back and find the note in your history, which is searchable across both the transcript and the processed output on web, iPhone, iPad and Mac. Translated notes carry a badge.

Transcribe a long audio file: the 20-minute and 2-hour lines

Three bands decide what happens to a recording, and length is the only thing that decides which band you are in. Up to 20 minutes, the file is processed immediately on every plan, guests included. From 20 minutes to 2 hours, it is processed in the background and Pro is required. Over 2 hours, it is rejected.

The 2-hour cap applies per recording, not per day or per month, so a three-hour lecture has to be split into two files before either part will run. Most recording and editing apps can export a section, and splitting has a second benefit: it helps with the limit that catches people out more often than duration does.

That limit is file size. The cap is 64 MB, whatever the duration. A 90-minute conversation exported as uncompressed WAV will often exceed it, while the same audio as MP3 or M4A is far smaller. Accepted formats are WAV, MP3, M4A, MP4, WebM, OGG and Opus, so re-exporting to a compressed format is usually enough to get a long recording through.

How long does it take to transcribe 1 hour of audio?

There is no single number for this, and quoting one would not help you, because it varies with the recording. What is predictable is the shape of the wait, and that shape changes at the 20-minute mark.

Under 20 minutes, you stay on the page and the note comes back to you while you wait. Over 20 minutes, the recording is queued on purpose, so that you are not left watching a progress bar for a job of that size. You close the app, do something else, and the finished note is in your history when you come back. Because history is tied to your account, you can start a 1-hour file on the Mac app and read the result on your phone.

A network connection is required while the audio is uploaded, since processing happens in the cloud rather than on your device. Once the audio has been received, though, nothing has to stay open for the work to finish, which is the whole point of handling long files this way.

Why a two-hour summary should not fade out at the end

The common failure mode with long audio is a summary that is detailed for the first twenty minutes and vague for everything after that. It happens when a two-hour transcript is squeezed through a single pass and the later material gets compressed down to a sentence or two.

sAInote handles long recordings hierarchically instead. The recording is broken into sections, each section is worked through on its own, and those results are built up into one note. The practical effect is that a decision made in the last twenty minutes of a meeting reaches the summary with the same weight as a point raised in the first five.

Long recordings also get speaker separation, which is what makes a two-hour interview readable rather than a wall of undifferentiated text. Special Names, the personal spelling glossary available on Pro, earns its keep at this length as well: a name spelled wrong in a five-minute note is a small irritation, but the same name across two hours can appear dozens of times. It is a spelling reference only, and never changes what was said.

What 360 Pro minutes a month covers

Pro includes 360 minutes of audio every month, which is 6 hours. Minutes are counted by duration, so a 90-minute recording costs 90 minutes: four recordings of that length use the full monthly balance. A full two-hour recording costs 120 minutes, so three of those fit into a month. A 3-minute voice memo costs 3 minutes in any mode.

The free tier is deliberately not built for this. Both the guest allowance and a free account give 15 minutes one time, not 15 minutes a month, and both are limited to Quick Note and Smart Summary. That is enough to see how the output reads, but a 1-hour recording will not run on it.

Pro is also what adds the other things that matter at length: Pro Assistant, where you write and save your own instruction and reuse it on any recording, Special Names, background processing, and priority support. An instruction can be written by hand, or drafted for you from a spoken description of what you want it to do, and each successful generation deducts 5 minutes from your balance whether or not you keep the result.

Long audio transcription: common questions

Can sAInote transcribe a 2 hour audio file?

Yes, 2 hours is the maximum length for a single recording. It requires Pro, because anything over 20 minutes is processed in the background rather than immediately. The file also has to be 64 MB or smaller, so at that duration a compressed format such as MP3 or M4A is safer than uncompressed WAV.

How long does it take to transcribe 1 hour of audio?

It varies by recording, and no fixed figure is published. The useful distinction is the shape: recordings up to 20 minutes come back to you while you wait, and a 1-hour recording is queued and processed in the background instead. You close the app, and the finished note is waiting in your history.

Can I transcribe a recording longer than 2 hours?

No, 2 hours is a hard cap per recording and longer files are rejected rather than truncated. Split the audio into parts under 2 hours each and process them separately. Each part counts against your minute balance by its own duration, so splitting costs you nothing extra.

Do I need Pro to transcribe a 90-minute recording?

Yes, anything longer than 20 minutes requires Pro. Guests and free accounts get 15 minutes one time, not monthly, which will not cover a 90-minute file at all. Pro includes 360 minutes a month, so a 90-minute recording uses 90 of them, or a quarter of the monthly balance.

What is the maximum file size for an upload?

64 MB, regardless of how long the recording is. Duration and file size are separate limits, and a long recording can fail the size check even though it is under 2 hours. Re-exporting to MP3, M4A, OGG or Opus instead of WAV is normally enough to get under the cap.

Can I close the app while a long recording is processing?

Yes, that is exactly what background processing is for. Once a recording over 20 minutes has been uploaded, you can close the browser tab or quit the iPhone, iPad or Mac app. The finished note appears in your history, which is tied to your account and available on every device you sign in on.

How many long recordings does Pro cover each month?

360 minutes of audio a month, which is 6 hours in total. That is three two-hour recordings, four 90-minute recordings, or six one-hour recordings, in whatever combination you need. Minutes are counted by duration, and a recording costs the same number of minutes in any processing mode.

Which audio file formats can I upload?

WAV, MP3, M4A, MP4, WebM, OGG and Opus. Anything in that list is accepted as long as it is under 2 hours and under 64 MB. For long recordings, the compressed formats are the practical choice, since an uncompressed file of an hour or more often breaks the 64 MB limit.

Does sAInote separate speakers in a long meeting recording?

Yes, long recordings get speaker separation, so a multi-person meeting or interview reads as distinct voices rather than one continuous block. It works on the recording after it is uploaded. There is no meeting bot, no live captioning during a call, and no connection to any conferencing service.

The rules, stated up front

If your file is under 2 hours and under 64 MB, sAInote will take it: immediately below 20 minutes, in the background above that on Pro. Anything longer needs splitting first, and it is better to know that before you upload.

Start a long recording

Related pages