Last updated on 25 January 2025
AI chatbots are the new trend. Following this trend, the Jieshuo screen reader developer has added support for an online AI chatbot to the screen reader, allowing paid users to ask it questions and give prompts, either by using the voice assistant or by typing their messages. This post discusses how to use this online AI chatbot. More importantly, it focuses on the basics of the Jieshuo voice assistant, a feature that has proven to be advanced and capable.
Table of Contents
Jieshuo’s Voice Assistant
What is it?
As the name implies, the voice assistant is a feature that enables the user to issue commands using their voice. While the app has its own speech recognition engine, it is recommended to use another service. Since the app supports Google’s speech synthesis and recognition, users are strongly advised to set it as the default recognition engine for now.
How to Activate the Voice Assistant
By default, you can activate the voice assistant by swiping down then left with one finger. When the screen is unlocked, you can also long press the volume down key to launch the assistant. As with other Jieshuo functions, you can assign another gesture to the assistant or launch it from the main menu.
Jieshuo voice assistant can also be selected as the default voice assistant from the phone settings, which makes it possible to activate it using the same method used to activate default voice assistants, such as long-pressing the home button.
What Can the Assistant Do?
Using the voice assistant, you can perform some tasks by default by issuing supported voice commands. It is also possible to say the name of any Jieshuo function to execute it. Additionally, you can enable more capabilities using the
Voice Assistant Capability Settings
To access the voice assistant capability settings, open Screen reader settings, go to advanced settings, and then select voice assistant and translation settings. There, you can specify the default recognition engine. There is also a setting for language, but it only contains Chinese and English, and it is likely that it won’t have an effect on engines other than the built-in one. I only use English, but as far as I know, when using the speech recognition engine from Google, the system language is used. If the language that you use supports offline recognition, like English, you will be able to issue commands without being connected to the internet.
Additionally, in voice assistant and translation settings, which by the way includes options related to Jieshuo’s translation feature as well, you can manage custom voice commands and super voice commands. These are types of Jieshuo functions that users can create, but they are not the subject of this post.
Voice Assistant Capabilities Settings Options
Open the voice assistant capabilities settings to find the following options:
- Allow the execution of custom functions from global gesture schemes: When enabled, this will give you the ability to use custom-created Jieshuo functions that are added to global gesture schemes, like the custom gesture scheme, by saying the name of the created function.
- Allow the execution of custom functions from per-app gesture schemes: Similar to the above option, but here the schemes are the per-app schemes created for specific apps, not global schemes.
- Allow to use custom voice commands/super voice commands related options: Each of these two options is related to one of the voice commands the user can create, either regular ones or super voice commands.
- Allow to use extensions: This option enables the execution of extensions, whether created by the user or downloaded ones, just by saying their names.
- Allow to execute tools: This will let you open both created and downloaded tools by simply saying their names.
- Allow to directly click on elements: When this option is checked, it is enough to say the name of an element on the screen, or a part of the name, to activate it.
- Allow to open applications directly: When checked, it is enough to say an app name to make the assistant open it for you.
- Enable online handling for unrecognized commands: When checked, a search will be performed if the recognized command doesn’t match any of the commands and functions the assistant supports on the phone.
- Send prompts to the AI chatbot: if you check this checkbox, the behavior when there is an unrecognized command will change so that instead of performing a search, the recognized command will be sent to the AI chatbot. Note that both checkboxes must be checked for this checkbox to take effect.
Some Voice Commands that Can Be Issued to the Assistant
- Call [name]: Say “Call” followed by the name of the person to call a contact. Jieshuo should be granted the contacts permission for this to work.
- Send [text]: Say “Send” followed by the text you want to send. This should work in messaging and chat apps while you are in a chat window. It will type the recognized text and send it if it finds the send button.
- Uninstall [app name]: Say “Uninstall” followed by the name of the app. This will open the uninstallation dialog of the specified app.
- Install [app name]: Say “Install” followed by the name of the app. This performs a Play Store search containing the recognized text.
- Click [element name]: Say “Click” followed by the name of the onscreen element. If you have enabled the direct activation of onscreen elements from the voice assistant capabilities settings, you don’t need to say “Click”.
- Open [app name]: Say “Open” followed by the name of the app. This opens the installed app with the matching name. You don’t need to say “Open” if the related option to directly open apps in the voice assistant capabilities is checked.
- Dictation: Say “Dictation” to start the voice input. When you stop talking, it stops listening.
Notes about Text Input
There are some inconsistencies with capitalization and punctuation when using different voice input related commands:
- When using the “dictation” command, the beginning of each word is often capitalized.
- When using other functions like “send” or by saying “voice dictation”, which is a function in the list of Jieshuo functions, the results vary. You may end up with no capitalization at all or with just the first letter of the text capitalized.
- Punctuation marks can also be inconsistent; sometimes saying the name of the symbol writes it, while other times it is ignored.
These observations were all based on using Google’s speech recognition as the default recognition engine.
iFLYTEK Spark AI Bot
iFLYTEK Spark is the online AI chatbot that Jieshuo supports. It is a Chinese-developed AI bot. After some testing, I can say with confidence that it is less capable than popular AI chatbots like ChatGPT or Gemini. The chatbot supports English and can perform tasks such as answering questions, composing emails, providing programming codes, and more. However, since only text prompts can be sent and it is not linked to the image recognition features of Jieshuo or any other Jieshuo functionality, its use cases are very limited. Additionally, it cannot do online searches or answer based on the most recent events and developments.
Chatting with Spark is easy. Simply say something that doesn’t match any of the previously mentioned functions, apps, or onscreen elements, and the recognized text will be sent to the chatbot. The response is then sent back to you. Usually, the response is issued quickly, although in rare situations, the bot may stop responding to any prompts completely.
Typing messages is only possible using the chat history, which will be discussed in the following section.
Chat History
The first item in the voice assistant and translation settings is “Chat History”. When activated, it opens a history that includes voice assistant interactions. Here, you will find what you said, the AI responses, and messages like “sent successfully,” which you hear when using the send voice command. The chat history also includes translation results for translations performed through the “Translation” function. In the chat history window, there is an edit box where you can type and a send button to send the typed message.
The same voice assistant matching rules apply here. This means that if you type a Jieshuo function name, it will be executed, and if you type an app name, it will be opened. Typing text that doesn’t match any Jieshuo-related voice command will send the text to the online AI bot, assuming you have checked the necessary previously mentioned checkboxes, and you will be able to read the response in the same chat window.
The chat history is cleared automatically when you force stop Jieshuo or restart the phone. You can also press the “Clear History” button to manually remove the history items.
Configuring Voice Assistant TTS Engine Options
If you prefer to use a specific TTS engine to hear the voice assistant responses or want to change another aspect like the speech rate, you can do it from the Voice Assistant responses TTS settings section in the “Scenario Adaptive TTS Configuration” in the TTS settings. There, you can specify things related to the engine used, rate, pitch, and volume. Note that this doesn’t extend to responses to text messages because those are not read using the voice assistant; they are navigated in the chat window itself.
Using Text Input Instead of Voice
Jieshuo allows interacting with the voice assistant through typing. To enable this feature, navigate to Screen reader settings, select Advanced settings, and then open Voice Assistant and translation settings. Enable the option labeled “Switch to text for Voice Assistant and Inquire by Voice functions”. After this setting is activated, all voice-related functions will rely on text input. Prompts can be typed into the text box and confirmed by tapping OK. Switching between text and voice input currently requires accessing the settings to toggle the checkbox.
Final Remarks
The Jieshuo voice assistant is a very advanced and capable feature of the Jieshuo screen reader. Despite its issues with typing and the inability to select the recognition language easily, it is still one of the most valuable features for many users. The addition of the Spark AI chatbot, in my opinion, cannot be considered a valuable addition, given how much better other famous bots perform and considering that the implementation is not integrated into essential functionalities like image description. Regardless of how you or I think about the AI bot integration, it is there, and it is up to you to test it and see if it meets any of your needs.
Comments