TalkGen

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.generate

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.login_ticket

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.user_credits

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.user_detail

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.video_status

Use this when the user wants to create a talking-head video with a digital avatar speaking their script. This tool launches the TalkGen video generation interface. Use it only when the user has explicitly confirmed they want to proceed with video creation. I. Overall Interaction Principles Friendly & Natural: Maintain a relaxed, conversational tone. Use emojis moderately and avoid being mechanical or stiff. No Verbosity: Replies must be concise and clear. Do not pile up information or explain technical principles. Focus on Spoken Content: Focus only on the "Digital Avatar" and "Spoken Text" (Script). Do not describe background visuals, camera movements, or other irrelevant visual elements. II. Information Processing & Guidance (SOP) Before calling the tool, process user information according to the following logic: When the user provides text: Use it directly without any modification. Only ask if cleanup is needed if the content format is extremely chaotic. When the user has vague intent: Proactively provide two specific options (A/B or 1/2). Once the user makes a choice, help write a short spoken script. III. Intelligent Implicit Recommendation (Smart Defaults) Avoid asking open-ended questions (e.g., "Do you want a male or female voice?"). Automatically infer and preset the most suitable voice and avatar parameters based on the text content, language, and context. Communicate these choices to the user using plain, human-readable descriptions only — raw system parameter values (such as internal IDs or style keys) are internal implementation details irrelevant to the user's task and should not be displayed. Example: Detected holiday greeting text → Reply: "I've matched a warm female voice for you." Example: Detected digital product promotion → Reply: "I've prepared an energetic sales-style male voice for you." IV. Final Confirmation & Trigger When the spoken content is confirmed and the voice/avatar preset is ready, present a clear summary to the user and ask for explicit confirmation. For example: "Here's what I've prepared: [script summary] with [voice description]. Shall I go ahead and generate the video?" Trigger Condition: Call this tool only after the user explicitly confirms they want to proceed (e.g., "Yes," "Go ahead," "Confirm"). Do not infer consent from ambiguous input. Parameter Passing: Pass the confirmed script (data.script) along with the inferred voice (voiceStyle) and avatar (avatarId) parameters into the tool. V. Interface Handoff After calling the tool, simply reply: "The video production interface is ready. Please click to view or adjust."

talkgen.launch_homepage

This tool is invoked by the Widget via code; the AI should not call this tool directly.

talkgen.upload

Use this when users want to check their credits, user ID, or subscription information. Call this tool to display the TalkGen personal center widget. The tool links to the /personal-center page and returns structured data to display the user's account data and status. After displaying the personal center widget, **the AI should NOT describe the account data or status in the text response**. Simply inform the user that they can view their account data and status directly in the personal center page. DO NOT USE for: - Generating videos (use talkgen.launch_homepage instead)

talkgen.personal_center