The four-hour ceiling is history. Upload audio or video recordings up to 20 hours and get back one clean transcript. Start the job, come back when it's done.
Some recordings refuse to be short. A full-day workshop, a board meeting that ran long, an eight-hour deposition, a weekend recording session. Until recently, FileToText met all of them with the same bad news: files past roughly four hours wouldn't process. If you work with long material, you know the ritual that followed. Cut the recording into chunks, transcribe each piece separately, then glue the text back together and hope nothing got lost at the seams.
Skip the splitting: our video to text tool takes a single recording up to 20 hours, first 10 minutes free.
We're happy to retire that ritual. The four-hour ceiling has been removed. You can now upload a single audio or video file and have the whole thing transcribed in one pass. We've verified the pipeline end to end on recordings of 10 hours.
The old limit wasn't a policy decision; it was a processing one. Very long files strained the way our pipeline held and worked through audio, so we capped what it would accept. That stage has now been rebuilt. Duration is no longer the constraint it used to be, and a nine-hour file moves through the same path as a nine-minute one: upload, transcribe, done.
In practice this means one thing: file length is no longer something you have to plan your work around.
The part we care most about is what you no longer have to do. No slicing the recording into parts. No queueing five uploads for one meeting. No pasting fragments together afterwards and re-reading the joins. You upload one file and receive one continuous, clean transcript: plain text you can read on the page, copy, or download.
The recordings people used to split are exactly the ones this release is for:
Here's the honest part of the announcement: ten hours of audio is a lot of audio, and transcribing it takes real time. A short clip still comes back in moments; a full-day recording takes noticeably longer, roughly in proportion to its length.
The right workflow is fire-and-forget. Kick the job off, close the laptop, and collect the transcript when you're back. Three things worth doing before you start a big one:
Nothing new to enable. It's the same flow as any other file:
If long files are a routine part of your work rather than an occasional event, the flow stays the same. Queue each recording as it lands, then collect the finished transcripts when you come back. Long jobs ask nothing extra of you beyond patience.
We've confirmed complete, clean transcripts on files up to 20 hours long. That comfortably covers full-day events, and for most users it means the length limit has effectively disappeared.
No. Splitting (and the stitching that came with it) is exactly what this update eliminates. One upload produces one uninterrupted transcript from the first word to the last.
Pay-as-you-go: $12, because one-off files cost $3 per started audio hour and are capped at $12 per file. On a subscription it uses 600 of your plan minutes, and minutes beyond your plan bill automatically at your plan's overage rate.
No. Accuracy holds steady across the whole recording; hour nine is treated no differently from minute one. What moves accuracy is source quality, which is what our accuracy benchmark measures: clear speech and a decent microphone matter far more than duration.
Got a monster file waiting? FileToText turns full-day events, marathon meetings, and multi-hour interviews into a single clean transcript. Upload your recording at filetotext.ai and let it run.