Hot News Voice Broadcast Generator

Problem Being Addressed

When producing news shorts or audio-visual updates, the hard part is often not finding material but assembling it: searching, filtering, rewriting, adding images, and generating voiceover can become a repetitive pipeline. This skill turns a hot-topic keyword into a cleaner broadcast script plus matching image and TTS audio, reducing manual iteration.

How It Works

The required input is news_hotspot, a specific news keyword. The skill first parses intent, extracts the news angle, and generates search keywords. It then fetches raw news material, cleans it, and extracts key entities, relationships, and a timeline. The result is rewritten into a standard news style and trimmed of filler phrases to produce a concise broadcast-ready text.

Image and audio generation run in parallel. One path builds an image prompt from the title, lead, and entities, then generates a news image. The other path sends the refined text to TTS to create a voice broadcast. Execution is asynchronous: a task is submitted and returns a task_id; the client must poll for status and feed intermediate results back to the user until text, image URL, and audio URL are complete.

Boundaries and Notes

It fits scenarios with a clear hot topic, such as tech, society, conferences, or breaking events, where the goal is fast production of a script, image, and voice asset. It is not a real-time fact-checking tool, and output quality depends on how specific the keyword is and how reliable the searched source material is. Without credentials it can run in simulation mode for preview only; real generation requires a ShowAPI AppKey and the async submit-and-poll workflow.