TL;DR: Wangyan OCR (望言OCR) is a Chinese open-source tool that does one job well — extracting hardcoded (burned-in) subtitles from video frames into an editable file. Its community edition is free and GPL-3.0 licensed; a professional tier (¥18/month or ¥88/year, roughly $12/year) unlocks faster processing, a Chinese-tuned recognition model, error flagging, and batch find-and-replace. It does not do speech recognition or translation. GeekLink also extracts hardcoded subtitles via OCR, but adds speech recognition (for audio-only sources with no existing subtitles), AI translation to 40+ languages, and a single batch pipeline across the whole workflow, all in a native macOS app. Choose Wangyan OCR if you only need low-cost Chinese-language OCR extraction. Choose GeekLink if you also need translation, speech recognition, or an end-to-end batch pipeline.
Need OCR extraction plus translation in one pipeline? GeekLink extracts hardcoded subtitles, transcribes speech, and translates to 40+ languages — locally on Mac, free to start.
Download FreeWangyan OCR (望言OCR, GitHub repo nhjydywd/SubtitleOCR) is a Chinese-market tool purpose-built for one job: reading subtitles that are already burned into a video's picture and turning them into an editable subtitle file. Unlike general-purpose transcription apps, Wangyan OCR does genuine optical character recognition on the video frames — it reads text that's part of the image, not audio.
The community edition is free and open source under GPL-3.0, with high-speed extraction, batch processing, and basic subtitle editing — a solid feature set for the core job. A paid "professional" tier costs ¥18/month or ¥88/year (roughly $12/year), which unlocks a faster processing mode (the developer's own marketing claims up to roughly 48x speed improvement over the free tier — that's their figure, not an independently verified benchmark), a proprietary recognition model tuned for Chinese text (including spacing conventions and traditional-character handling), an error-detection feature that flags likely OCR mistakes for review, project-wide find-and-replace history, and additional export formats.
Like a number of open-source tools before it, Wangyan OCR followed a fairly common playbook: the GPL-3.0 open version built an initial user base, and starting from version 1.4 a paid activation model was introduced alongside it. The freely available open-source version has seen less active development since. This isn't a knock on the project — it's just how the business model evolved, and it's worth knowing if you're deciding which edition to install.
What Wangyan OCR does not do is speech recognition or translation. It has no audio transcription capability at all — if a video has no existing on-screen subtitles, Wangyan OCR has nothing to extract, and it doesn't generate captions from spoken audio the way a transcription tool would. It also doesn't translate the subtitles it extracts into another language; its job stops at recovering the original on-screen text.
The comparison below covers the parts of the subtitle workflow each tool actually handles. Wangyan OCR is a focused, low-cost OCR extraction tool; GeekLink covers the same OCR job plus speech recognition, translation, and burn-in in one pipeline.
| Feature | GeekLink | Wangyan OCR |
|---|---|---|
| Platform | macOS (native, Apple Silicon optimized) | Windows-first open-source tool; Mac experience is not the primary target platform |
| Primary focus | Full subtitle production pipeline (recognition → OCR → translation → burn-in) | Hardcoded subtitle OCR extraction only |
| Video OCR (burned-in subtitle extraction) | Yes — handles colored text, outlines, busy backgrounds, ignores logos/watermarks | Yes — this is its core feature, with a Chinese-tuned recognition model in the paid tier |
| AI speech recognition | Yes — for audio-only sources with no existing subtitles, 100% local | No — purely image-based OCR, no audio transcription |
| AI translation | Yes — Claude 3.5 Haiku, GPT-4o, GPT-4o mini, DeepSeek (40+ languages) | No — no translation or cross-language localization |
| Subtitle burn-in | Yes — with font/color/position styling | Not part of the core workflow |
| Batch processing | Yes — 50+ videos across the whole pipeline (OCR + speech + translation + burn-in) in one run | Yes — batch OCR extraction across multiple videos |
| Error / low-confidence flagging | Yes — surfaces likely misreads for review | Yes — paid tier flags likely OCR mistakes |
| License / openness | Closed-source commercial app with a free tier | Community edition free and GPL-3.0 open source; paid tier proprietary |
| Local processing / privacy | Speech recognition, OCR, and burn-in run 100% locally; only subtitle text (not video) goes to the LLM for optional translation | OCR runs locally (open-source codebase) |
Key takeaway: both tools genuinely do OCR extraction of hardcoded subtitles — that overlap is real. The difference is scope. Wangyan OCR stops once it has extracted the on-screen text. GeekLink treats extraction as one step in a longer pipeline that can also transcribe audio-only sources and translate the result into another language, all in the same batch run.
Wangyan OCR's community edition is free and open source (GPL-3.0), with high-speed extraction, batch processing, and subtitle editing included at no cost. The professional tier costs ¥18/month or ¥88/year (roughly $12/year), which is a genuinely low price — it reflects a Chinese consumer software market where prices are typically much lower than Western SaaS pricing, not a lesser product for that specific job.
GeekLink offers a permanent free tier (full speech recognition, OCR, editing, batch processing, SRT/ASS export; free exports carry a small GeekLink credit). The paid plans are:
These are different price points for different jobs, not a "cheap vs expensive" comparison of the same product. Wangyan OCR's ¥88/year professional tier is priced for a narrow task — faster Chinese-language OCR extraction — inside a low-ARPU consumer software market. GeekLink's $99/year or $169 lifetime price is for a broader, cross-language pipeline aimed at a different buyer: someone who needs OCR, speech recognition, AI translation, and burn-in to all work together, and who values that breadth enough to pay accordingly. If all you need is Chinese OCR extraction, Wangyan OCR's price is hard to beat for that specific job.
Wangyan OCR is a genuinely good fit in a few specific situations:
You only need Chinese-language hardcoded subtitle extraction. If every video you work with already has burned-in Chinese subtitles and your job ends once you have that text as an editable file — no translation, no cross-language work — Wangyan OCR's recognition model is tuned specifically for Chinese text, including spacing conventions and traditional-character handling.
You want the lowest possible cost for OCR-only extraction. At ¥18/month or ¥88/year (roughly $12/year), the professional tier is inexpensive, and the community edition is free and open source. If your workflow is purely "get the burned-in text out," there's no need to pay for a broader pipeline you won't use.
You want an open-source codebase. The GPL-3.0 community edition means you can inspect, modify, or self-host the extraction logic — something a closed-source commercial app can't offer.
You already have separate tools for translation and video editing. If your pipeline already stitches together an OCR tool, a translation service, and an editor, Wangyan OCR can slot in as the OCR step without paying for a bundled alternative you don't need.
GeekLink is the stronger choice when your job extends beyond OCR-only extraction of Chinese-language video:
You need translation. This is the clearest difference. Wangyan OCR has no translation feature at all — it extracts the original on-screen text and stops there. GeekLink translates extracted (or transcribed) subtitles into 40+ languages using Claude 3.5 Haiku, GPT-4o, GPT-4o mini, or DeepSeek, reading surrounding lines for context. If your audience needs the content in a different language, Wangyan OCR simply can't take you there.
You also work with audio-only sources that have no existing subtitles. Wangyan OCR only reads text that's already burned into the frame — if a video has no on-screen subtitles, there's nothing for it to extract. GeekLink's speech recognition covers that case by transcribing the spoken audio directly, so the same app handles both "video already has hardcoded subtitles" and "video has no subtitles at all."
You want one batch pipeline instead of stitching tools together. Wangyan OCR batches the OCR step. GeekLink batches the entire workflow — OCR or speech recognition, translation, editing, and burn-in — across 50+ videos in a single unattended run, so you're not manually passing files between an OCR tool, a translation tool, and an editor.
You want a native macOS app. Wangyan OCR is a Windows-first open-source tool; its Mac experience is not the primary target platform and may be rougher as a result. GeekLink is built only for macOS and optimized for Apple Silicon (M1/M2/M3/M4).
You're serving a non-Chinese-speaking audience. Wangyan OCR's recognition model is tuned specifically for Chinese text. If your source videos have subtitles in other languages, or you need output in other languages, GeekLink's broader OCR and translation range is the better fit.
No. Wangyan OCR does not offer translation or any cross-language localization feature. It is purely an OCR/extraction tool: it reads text that's burned into the video frames and outputs it as a subtitle file in the language it was originally written in. Its paid tier adds speed, a Chinese-tuned recognition model, error flagging, and export format options — but none of these touch translation.
If you extract Chinese hardcoded subtitles with Wangyan OCR and then need them in English, Spanish, Japanese, or any other language, you'd need to hand the extracted file to a separate translation tool or service — Wangyan OCR's job ends at recovering the original text.
This is where GeekLink's OCR and translation being part of the same pipeline matters. GeekLink extracts hardcoded subtitles the same way Wangyan OCR does, but the extracted text then flows directly into AI translation (Claude 3.5 Haiku, GPT-4o, GPT-4o mini, or DeepSeek) without leaving the app, and the translated result can be exported as a subtitle file or burned back into the video. For anyone whose subtitles need to reach an audience in a different language — a Chinese short drama being localized for a Western audience, for example — that's a meaningfully different workflow than extraction alone.
The community edition is free and open source under GPL-3.0, with high-speed extraction, batch processing, and subtitle editing included. A paid professional tier costs ¥18/month or ¥88/year (roughly $12/year) and unlocks faster processing, a Chinese-tuned recognition model, error flagging, batch find-and-replace, and more export formats.
No. Wangyan OCR is purely an OCR extraction tool — it reads text burned into video frames and outputs it in the original language, with no translation or cross-language localization feature. GeekLink extracts hardcoded subtitles the same way and then translates them into 40+ languages in the same pipeline.
No. Wangyan OCR is purely image-based OCR of on-screen text — it has no audio transcription capability. If a video has no existing hardcoded subtitles, there's nothing for it to extract. GeekLink adds speech recognition for exactly that case, transcribing spoken audio directly when a video has no on-screen subtitles at all.
Wangyan OCR is a Windows-first open-source tool; its Mac experience is not the primary target platform. GeekLink is a macOS-native app that extracts hardcoded subtitles from video (the same core job as Wangyan OCR), and additionally offers speech recognition for audio-only sources, AI translation to 40+ languages, and batch processing across the full pipeline — all optimized for Apple Silicon.
If you only need low-cost, Chinese-language hardcoded subtitle extraction with nothing else, Wangyan OCR is a solid, purpose-built, inexpensive option. Choose GeekLink if you also need translation, speech recognition for audio-only sources, or want OCR, transcription, translation, and burn-in to run together in a single Mac-native batch pipeline.
Mac subtitle factory with hardcoded-subtitle OCR, AI transcription, AI translation, and burn-in. Free tier available — no account required.
Free DownloadDisclosure: GeekLink is our product. Wangyan OCR information is based on its public GitHub repository (nhjydywd/SubtitleOCR) and publicly listed product/pricing information as of this article's publish date; features and pricing may have changed since — verify current details on the project's official page.