Unlimited OCR on your device: read manga images into editable text
Published · Last updated
Part of the guide: On-Device Voice Guide
Syllivo's unlimited OCR turns manga and comic images into text you can edit and send to Studio, entirely on your device. There is no per-image quota and no per-use quota: you can read as many images as you like, one after another, and the practical limits are your device's memory, storage, and time — not an allowance that runs out. When you are ready to try it, open Studio and then open the Transcript tab, where the Manga images mode lives. If you also want spoken words turned into text, read unlimited transcription on your device as well.
What unlimited OCR is, and what it is not
Unlimited OCR in Syllivo means the reading itself is never metered. There is no set batch limit: images are processed one by one, in the order you added them, and finished images are kept even if the device runs out of memory partway through a long queue. Only the recognised text is stored. The original images are never stored, so reading a page does not fill your device with copies of it.
It does not mean every page will read perfectly, and it does not mean the result is a translation. The text comes back in the image's own language, in editable blocks — one block per speech bubble or caption — and you edit it before sending it anywhere. Nothing is guessed to fill a gap: if a bubble cannot be read, it comes back empty rather than invented.
Pick the reader that fits your page
The Transcript tab offers three readers, and choosing the one that fits your page is the single most useful decision you will make:
- General (current) — the default reader, and the best all-round choice for English and Chinese pages.
- New reader — a newer reader for English and Chinese pages, worth trying when the default reader misses small or faint text.
- Manga specialised — the reader built for Japanese manga pages; Chinese pages are also covered, but English is not promised for this reader.
Language coverage, stated plainly: English and Chinese are covered, in both Simplified and Traditional forms, and Japanese is covered by the manga reader. Hindi is not supported for image reading in this version. A Hindi page will always come back empty here, no matter which reader you pick — that is a limit of the readers, not a problem with your image, and no clearer photo will change it.
How to use it well
A few habits make a large difference, and they cost nothing:
- Set the reading order before you read. Manhua left-to-right is the default; Manga right-to-left reverses the block order for Japanese-style pages. A page read in the wrong order is not misread — it is just in the wrong order — so set it first, or use the per-image override and read that image again.
- Start with clear images. A sharp, well-lit page with readable lettering beats a blurry screenshot every time, and a short test on one page tells you more than a long queue run on a guess.
- If one reader misses text, try another reader on the same image. The readers fail in different places: what one misses, another may catch, and re-reading costs nothing but time.
- Use the fallbacks the app already runs. When a first pass finds nothing, the readers try further passes for the hard cases: inverted text (light lettering on a dark panel), small text on a full page, and artwork-heavy pages where text is scattered among drawings. Tall pages are handled by tiling, so a very tall page is read tile by tile instead of being refused.
- Work from the text, not from the image. Each image gives you editable blocks with its own Copy and Send to Studio actions, and the batch actions copy, save, or send only the finished images you selected, in queue order. Send to Studio always starts a new project; it never overwrites one.
Honest limits
Unlimited here means no quota, not no limits. The limits that actually apply are these: your device's memory and storage decide how many images you can queue comfortably; a long queue takes a long time, roughly in proportion to the number of images; and stylised or handwritten lettering remains a weak case for every reader, so a decorative font may still come back empty. If a page fails on all three readers, the honest answer is that this page, in this version, cannot be read — try a clearer image, or accept that this particular lettering is beyond the readers for now.
Frequently asked questions
Is there really no limit on how many images I can read?
There is no per-image or per-use quota and no set batch limit. The practical limits are your device's memory, storage, and time, and finished images are kept if the device runs out of memory partway through a queue.
Which reader should I use for an English manhua page?
Start with General, the default reader — it is the best all-round choice for English and Chinese pages. If it misses small or faint text, try the New reader on the same image before changing anything else.
Why does a Hindi page always come back empty?
Hindi is not supported for image reading in this version, in any reader. A Hindi page will always come back empty here, no matter how clear the image is.
Are my images uploaded or stored?
No. Reading runs on your device, and only the recognised text is stored. The original images are never stored.
What happens if the device runs out of memory during a long queue?
Finished images are kept. Only the image in progress is lost, and you can read the remaining images again.