Unlimited transcription on your device: audio and video into editable text
Published · Last updated
Part of the guide: On-Device Voice Guide
Syllivo's unlimited transcription turns spoken words into timestamped text you can edit and send to Studio, entirely on your device. There is no per-transcription quota: you are not charged per minute, per file, or per use, and nothing is uploaded to a transcription server. When you are ready to try it, open Studio and then open the Transcript tab, where the Audio mode lives. If you want text out of pictures instead, read unlimited OCR on your device as well.
What unlimited transcription is, and what it is not
Unlimited transcription in Syllivo covers two kinds of input, and they do not share the same limit — so the honest definition matters:
- Video has no length limit. A video of any length can be transcribed: its audio track is extracted on your device, and the audio is transcribed in short, 30-second windows, in order. Progress shows as "Transcribing minute X of Y" while it works, and stopping partway keeps every finished segment — a stopped transcription is a partial result, never a failure.
- Audio has an explicit 5-minute limit in this version. An audio clip longer than 5 minutes is refused outright; it is never silently cut short. If your recording is longer than that, split it into shorter clips first, or use the video route for long material.
So "unlimited" here means no per-use quota and no length limit on video — not "every input, of any length, in every mode." The practical limits that remain are your device's time and memory: a long video takes a long time, roughly in proportion to its length and slower on a weaker device, and the tab must stay open while it works, because transcription does not continue after the browser closes.
Audio mode: record or upload
Audio mode accepts a recording made in the app or an audio file you upload. The language choice is Auto-detect, Hindi, or English, and Hinglish is best-effort auto-detect: a speaker who switches between Hindi and English mid-sentence is the weakest case, so check the script before use rather than trusting it blind.
A recording does not need to be studio-clean, but it does need to be understandable: one speaker at a time, in a reasonably quiet place, beats a long, noisy recording. If no speech is found, the result says so plainly instead of inventing words.
Video mode: the audio track, transcribed
Video mode accepts MP4, M4A, and MOV videos whose audio track uses the AAC format — the formats a phone camera produces. Other video formats are refused with a plain message listing the supported ones; they are never silently converted. A video with no audio track is refused plainly too, because there is nothing to transcribe.
The video file itself is never uploaded and never stored: only the audio track is extracted, on your device, and only the transcribed text is saved. The same language choice applies as in audio mode, and the output is the same shape: timestamped, editable segments you can copy, save, or send to Studio as a new project — never overwriting an existing one.
How to use it well
- Test short first. Transcribe one short clip and check the result before running a long video. A short test tells you whether the language choice, the recording quality, and the speaker's pace are working, before you spend a long time on a full run.
- Split long audio. Because audio mode refuses anything over 5 minutes, cut a long recording into shorter clips first rather than discovering the refusal at the end of an upload.
- Keep the tab open for long videos. Transcription stops if the browser closes, so start a long video when you can leave the tab open — and remember that stopping deliberately is safe, because finished segments are kept.
- Edit before you send. Segments are editable for a reason: fix names, numbers, and any mid-sentence language switch, then send the cleaned text to Studio as a new project.
Honest limits
Beyond the 5-minute audio limit, the limits are practical ones: device time and memory, a tab that must stay open, and the usual transcription weaknesses — heavy background noise, overlapping speakers, and mid-sentence Hinglish switching all reduce accuracy. A transcript is a starting point you edit, not a finished script, and a stopped run keeps what it finished rather than pretending it finished everything.
Frequently asked questions
Is transcription really unlimited?
There is no per-transcription quota. Video has no length limit at all, but audio mode has an explicit 5-minute limit in this version, and the practical limits for both are your device's time and memory.
What video formats work?
MP4, M4A, and MOV videos with an AAC audio track. Other formats are refused with a plain message, and a video with no audio track is refused because there is nothing to transcribe.
What happens if I stop a long video transcription halfway?
Finished segments are kept, in order, as a partial result. Stopping is never reported as a failure, and you can edit and use what was completed.
Is my audio or video uploaded anywhere?
No. The audio track is extracted and transcribed on your device, and only the transcribed text is saved. The video and audio files themselves are not stored.
Why was my 8-minute audio file refused instead of cut down?
Audio mode refuses clips longer than 5 minutes rather than silently truncating them, so you always know exactly what was and was not transcribed. Split the recording into shorter clips and transcribe those instead.