Jieshuo (formerly known as Commentary Screen Reader or CSR), the popular Android screen reader for the visually impaired, has rolled out its latest update: version 20241108. This release introduces a series of improvements and new features, with a major focus on enhancing offline OCR (Optical Character Recognition) capabilities.
Table of Contents
- 1. Major Upgrade to Offline OCR: More Accurate and Lighter
- 2. Blue Hart LLM Integration for AI-driven Features
- 3. Video Description for Premium Users: A Work in Progress
- 4. Enhanced Volume Controls
- 5. New Addition to the List Copying Function
- 6. Answering and Ending Call Gestures/Functions Should Work Again with WhatsApp Calls
- Download the Latest Version
1. Major Upgrade to Offline OCR: More Accurate and Lighter
The hallmark of this release is the complete revamp of the offline OCR engine, which now supports English, Chinese, Japanese, Korean, Russian, German, French, Italian, Portuguese, and Spanish, whereas the old engine only supported English and Chinese.
Based on tests conducted by the Accessible Android editorial team, the updated offline OCR engine is more accurate. Additionally, the optimized engine has resulted in a reduction in the APK installer size, bringing it down to just 58MB. This streamlined package ensures that users can enjoy enhanced OCR performance without taking up excessive storage space.
2. Blue Hart LLM Integration for AI-driven Features
Another noticeable change in the new release is the replacement of the old AI LLM by the BlueHeart LLM, developed by Vivo. The new AI model powers the picture description, icon recognition, face recognition, and inquire by voice functionalities. It is also free from the 100-recognitions-per-day limit that was in place in previous versions using the old AI model.
3. Video Description for Premium Users: A Work in Progress
Jieshuo has introduced a video description feature exclusively for premium users. This new addition allows users to extract information from videos, which can be activated through a custom gesture or the expanded main menu, or from the recognition menu.
However, based on initial tests, this feature appears to be more of a continuous detailed image description function rather than a full-fledged video description tool. It’s not yet on par with more advanced video description solutions like Seeing AI or PiccyBot, which provide richer contextual information. As such, it may require further refinement to reach its full potential.
4. Enhanced Volume Controls
This update also includes tweaks to volume control responsiveness, allowing for smoother adjustments. Additionally, when users reach maximum (100%) or minimum (0%) volume levels, Jieshuo will now continuously announce “100” or “0” if further adjustments are attempted, providing clearer audio feedback.
5. New Addition to the List Copying Function
A useful enhancement to the list copy function has been added, enabling users to either save results as a local text file or copy them to the clipboard. This offers greater flexibility for managing extracted text and data.
6. Answering and Ending Call Gestures/Functions Should Work Again with WhatsApp Calls
In the new version, an attempt has been made to fix the issue preventing users from answering or ending WhatsApp calls using Jieshuo’s answering and ending call gestures and functions. The developer is consistently working to ensure that WhatsApp calls are included in the list of calls manageable through these functions, a feature that occasionally encounters issues.
Download the Latest Version
To download the latest version of Jieshuo, click the link below:
With its improved offline OCR, AI engine updates, and new features like video recognition, Jieshuo continues to empower its users by providing a robust and efficient screen reader experience. While some features, like video recognition, are still in development, this release demonstrates Jieshuo’s commitment to enhancing accessibility for the visually impaired community.
Stay tuned for more updates as Jieshuo evolves to meet user needs.
I purchased a year of this thing back in the summer thinking I wouldn’t use it much. Within 48 hours I had switched to using it very nearly full-time. I will only switch back to talkback when I have to post a podcast episode because my hosting provider has refused to fix their description edit box to a way that will allow for The voice dictation function to work. Not that it’s an issue in this case because there’s no Braille support, but that type of box also doesn’t let you type with a Braille display. These new features are even more incentive than I already had to pay for more years. When I get the chance, I may end up getting the two or three year package.