WonderFlow: Narration-Centric Design of Animated Data Videos

TL;DR

WonderFlow addresses the challenge of producing narrated data videos by unifying text-visual linking with a structure-aware animation library and TTS-based narration synthesis in a single authoring tool. The approach allows authors to map narrative segments to chart elements, automatically generate audio, and synchronize animations on a shared timeline, enabling real-time previews and iterative refinement. Key contributions include a formative study, the narration-centric WonderFlow tool, a structure-aware animation library, an end-to-end design pipeline, and comprehensive evaluation via an example gallery, a novice user study, expert interviews, and a comparison with PowerPoint. Results indicate that WonderFlow is easy to use, reduces interaction effort, and supports expressive narration-animation interplay, offering a practical pathway for accessible data storytelling. Future work envisions broader domain support, deeper AI-assisted automation, and integration with business intelligence tools to scale narration-driven data communication.

Abstract

Creating an animated data video enriched with audio narration takes a significant amount of time and effort and requires expertise. Users not only need to design complex animations, but also turn written text scripts into audio narrations and synchronize visual changes with the narrations. This paper presents WonderFlow, an interactive authoring tool, that facilitates narration-centric design of animated data videos. WonderFlow allows authors to easily specify a semantic link between text and the corresponding chart elements. Then it automatically generates audio narration by leveraging text-to-speech techniques and aligns the narration with an animation. WonderFlow provides a visualization structure-aware animation library designed to ease chart animation creation, enabling authors to apply pre-designed animation effects to common visualization components. It also allows authors to preview and iteratively refine their data videos in a unified system, without having to switch between different creation tools. To evaluate WonderFlow's effectiveness and usability, we created an example gallery and conducted a user study and expert interviews. The results demonstrated that WonderFlow is easy to use and simplifies the creation of data videos with narration-animation interplay.

Figures (8)

• Figure 1: The pipeline of narration-centric design of animated data videos.
• Figure 2: User interface of WonderFlow. Users can first select the text phrases in the narration editor (a) and visual elements from the canvas (b) to form text-visual links. Then they can apply an animation preset selected in the animation effect panel (c) to the visual elements. WonderFlow then generates a narration-animation pack on the timeline (d).
• Figure 3: Interactions of WonderFlow. The user can input the narration and WonderFlow automatically generates the narration audio with timestamps (a). Users should first select narration words on the audio timeline (a) and then select visual elements on the canvas with various selection modes (b). Next, users can preview the filtered animation presets in the animation effect panel and apply an appropriate animation effect to the text-visual link (c). After that, hovering on the animation icon will show a preview of the created narration-animation binding (d). Then, users can click the "Play" button to compile the video for iterative preview (e). Finally, users can fine-tune the timeline ( e.g., duration and start) after preview (f) and iteratively design subsequent animations.
• Figure 4: The generated data video in the use case. The bottom part shows the narration-animation linkings; the upper part is the video frames in specific timestamps, marked on the narration. The animation effects are as follows: at first, the canvas only presents a title, then the axes float in, followed by bars growing upwards from the bottom. Next, the annotation "976200" above the bar fades in. Finally, the three groups of annotations representing three major increases zoom in first and then zoom out in order (The data video can be found at https://datavideos.github.io/WonderFlow/).
• Figure 5: Structure-aware animation presets. The upper part is three structure-aware animations created in the usage scenario (Figure \ref{['fig:use case']}). The middle part is the internal structure of the presets. The square bracket ([]) in the animation card indicates soft constraints, which means that only if the elements exist, the corresponding animation effect will be applied. The asterisk (*) indicates that the animation can be applied to any visual element. The bottom part is the animation preview on the animation effect panel (Figure \ref{['fig:user interface']}-c).