Dijk van een Wijf: Speaking with the Wad

An immersive journey through nature, guided by voice, sound and AI

Introduction

For the artwork Dijk van een Wijf, created as part of Sense of Place, we were invited to create something that goes beyond a traditional audiotour.

Instead of telling visitors what they are looking at, the idea was to let them engage in a conversation with the landscape itself.

Not a fixed story.
Not a linear experience.

But something dynamic, responsive and alive.

Speaking with the Wad

At the heart of this project lies a simple but powerful idea: 

What if nature could speak back?

Visitors can ask questions about the Wadden landscape, the artwork, the history, the animals and everything that exists in that space. 
The system responds in real time, not as a guide, but as an entity, a voice that feels like it belongs to the land itself. This required a careful balance. The responses needed to be informative and grounded in real knowledge, but also poetic, slightly abstract, and at times almost otherworldly. The goal was not just to inform, but to evoke a sense of wonder.

The artist

To make this believable, everything started with the voice.

We collaborated with Hiske, a Frisian artist with a strong background in music and theatre. Her ability to shape tone, rhythm and emotional nuance made her the perfect voice for this project. Just as important was her connection to the region, which subtly carries through in the delivery.

We recorded hours of material in both Dutch and English, carefully directing performance, pacing and articulation. The intention was to capture a voice that feels calm, grounded and human, yet not entirely human. Something that sits between a person and a natural force.

The voice of the Wad

Recording with intention:

Every recording session was approached with a very specific tone in mind. We explored pacing, silence, softness and breath to create a sense of space within the voice. The scripts were designed to cover a wide range of phonetics, while maintaining a consistent emotional layer of curiosity and calm.

Building the voice model:

From these recordings, we selected the strongest material and trained a voice model. This required extremely clean input, so we carefully edited out noise, mouth sounds and inconsistencies. The cleaner the dataset, the more natural and stable the generated voice becomes.

Consistency across languages:

One of the challenges was maintaining the same identity in both Dutch and English. Through careful direction, processing and model training, we ensured that the voice feels like the same entity, regardless of language.

Seasonal sound design

Musical modes as emotional tools

We used musical modes as a framework to guide the emotional tone of each season. Minor tonalities create a sense of depth and stillness in winter, while lydian and major tonalities introduce lightness and openness in the warmer seasons.

From nature to music

From these recordings, we selected the strongest material and trained a voice model. This required extremely clean input, so we carefully edited out noise, mouth sounds and inconsistencies. The cleaner the dataset, the more natural and stable the generated voice becomes.

Designed for conversation

A key challenge was making sure the sound never competes with the voice. The soundscape supports the conversation, breathes with it, and leaves space for the interaction to remain clear and intimate.

System & interaction design

 

Behind the scenes, a carefully designed system connects voice, knowledge and interaction into one seamless experience.

  • Knowledge-driven responses

    The agent is connected to a structured knowledge base covering ecology, history, the artwork and the environment. This ensures that every response is meaningful and relevant.

  • Shaping the spoken interaction

    A big part of the experience was making sure the generated responses did not feel mechanical once spoken aloud. By fine-tuning rhythm, timing and tone, we created a vocal flow that feels more like a presence in the landscape than a piece of technology.

     

  • Application and interface

    The experience is accessed through a custom interface, designed for on-location use. With a simple interaction, visitors can enter the conversation and explore at their own pace.

  • Continuous testing and refinement

    We iterated extensively, adjusting everything from response length to audio balance. Small changes in timing or tone had a big impact on how believable the experience felt.

The result

The final result is an experience where technology disappears into the background.

What remains is a quiet, immersive interaction.

You stand in the landscape, ask a question, and something answers, not as a device, but as part of the environment itself.

It invites people to slow down, to listen, and to engage with the space in a different way.

Speak with the wad

Interested in creating something similar?

We’re always open to exploring new ideas.

Whether it starts with a question, a concept or a location, we’re happy to think along and shape it into an experience that feels as natural as it sounds.

Contact