Processing times | Speak AI Docs

Speak AI processes files far faster than real time. A 30-minute meeting is usually ready in under a minute, a 1-hour file in a minute or two.

Speak AI processes files far faster than real time. As a rule, expect roughly a minute or two of processing for every hour of recording:

A 10-minute recording is ready in well under a minute
A 30-minute meeting takes under a minute
A 1-hour file takes a minute or two
A 2-hour recording takes around three minutes

Around 9 in 10 files finish within five minutes. Processing time scales with length, so a longer file takes longer, but even a 3-hour recording usually comes back in under 10 minutes.

Very short files have a small fixed startup cost, so a two-minute clip and a ten-minute clip finish in a similar amount of time. Files longer than 4 hours are split into segments and transcribed in order, which takes longer than the rate above suggests. See Duration limits.

What affects the speed

File length: longer files take proportionally longer, at roughly the rate above.
File format: standard formats such as MP3, MP4 and WAV go straight through. An unusual format is converted first, which adds time. See Supported formats.
Upload speed: your file has to reach Speak AI before processing starts, so a slow connection adds to the total wait on a large file.
Large batches: when you upload many files at once, Speak AI releases them steadily rather than all at once. The last file in a big batch starts later than the first.
Server load: processing can run slightly longer at peak times.

When a file takes longer than expected

Track progress from your dashboard, where a spinner means the file is still working and a green check means it is done. Speak AI also emails you when the file is ready. See Status for what each state means.

If a file is still processing after 30 minutes, whatever its length, something has gone wrong:

Refresh the page in case the status has already updated.
Re-transcribe the file from its options menu.
If it fails again, send us the file name and we will investigate. See Transcription troubleshooting.

Evaluating Speak AI for a team? Book a demo and see your own recordings analyzed.

Related: Transcription · Accuracy