#2026-M06 --- ## [Sound & Image Lab](https://keiaiendiel.github.io/soundviz-lab/) I started a wider piece of research into how sound can be turned into image. It is a problem that has interested me since my very first interactive project, [Zvuk.](https://kindl.work/ZVUK) I went back to some of the oldest inspirations and ways a sound signal can be carried over into something visual. To keep it from being arbitrary, I tried to collect approaches grounded in real physical phenomena, such as *Faraday waves on a liquid surface, Chladni figures, photoelastic refraction, the optics of breath, the oscillographic trace of a signal,* and more. I put it all on the [Sound & Image Lab](https://keiaiendiel.github.io/soundviz-lab/) site, where you will find over fifty such approaches side by side, from a clean line through texture to a full surface. Everything should react to a live audio input once you allow it in the browser. Sound is not only about loudness; the signal itself carries far more useful data, *whether it is a clean tone or noise, its pitch, when a hit lands, rhythm and onsets.* The program also reads the shape of the waveform itself, so with the oscillographic drawing, for instance, what you see is literally what the microphone just picked up. ![[soundimage.gif]] ![[sound-image.jpg]] --- ## Tykadla / Antennae [Last month](https://kindl.work/Projects/Newsletter/2026-M05) I tried to sketch three basic positions an *"electric organism"* can take. The **Visitor** that gets to know a place, the **Inhabitant** that just persists, and the **Translator** that interprets. I am now mostly building the third of them. The object carries the working name **Tykadla** *(Antennae)* and is a static organism meant to stand directly in the landscape. Each unit is 1.5 to 2 m tall and is made of two vertical light-guiding antennae rising from a low body. It also has notional **sensory roots,** interchangeable probes that lend the machine a sensory reach into the surrounding ecosystem. *It reads bioacoustics in the soil, contact vibration on nearby surfaces, humidity and light, and possibly chemical or spectral inputs.* *Antennae* work as a *tool of translation.* They pick up *what happens at frequencies and on timescales people cannot perceive directly,* and carry it over into a slow channel a visitor can notice. Most of the time the unit is quiet. It logs what it hears to a local archive and shows only a faint sign of life. A small radar in the body reads the surroundings, and once someone comes close the channel opens: the object brightens and starts to sound according to what the sensory roots are picking up at that moment. So it speaks to the outside *only when there is someone to hear it.* The object follows directly from two earlier works ([Receptor](https://kindl.work/2023/Receptor) and [Synapse](https://kindl.work/2023/Synapse)) and pushes that line out into the landscape. ![[synapse-wide.jpg]] ### Sensing Again I am building it all as a platform, so the final *perceptual system* will depend on a lot of experiments straight out in the field. *If something occurs to you that would be interesting to watch outdoors like this, get in touch.* For now I have split it, on paper, into 3 basic layers; | | | | | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Below the surface.** A geophone, a microphone made to pick up vibration straight in the ground, to record the full soundscape under the surface. Whether it is the life of underground organisms or the vibrations that sound down there. | **On the surface.** A piezo disc, fixed straight onto the bark of a tree or its roots, or other forest growth. Over a long span the object can follow a tree coming apart from inside, wood swelling, shrinking or cracking, and any signs of other organisms. | **In the air.** Humidity, temperature and light. A very accessible signal and data that shift dynamically over time and can steer the object's behaviour. On top of that I want to use the radar to read the surroundings, so it knows when a person (or maybe a larger animal) comes close to the *antennae.* | ### Hardware ``` || || || 2x light-guiding || amber, glows full length || rod (antennae) || = internal state \\ // \\ // +------------------------------+ | ESP32-C3 / Teensy (brain) | | 24 GHz radar presence | <- a few metres | amber LED light | two outward channels: | speaker sound | light + sound | microSD archive | logs even while quiet +------------------------------+ ^ solar 5V -> LiFePO4 charger -> 4 cells | sensory roots / connectors | _______________________________________________________ | | | +--------------+ +--------------+ +---------------------+ | GEOPHONE | | PIEZO | | CLIMATE | | (soil) | | (bark) | | humidity/temp/ | +--------------+ +--------------+ | light (air) | a hand's depth onto bark +---------------------+ in topsoil wind + chewing damp and dry ``` The brain is an *ESP32-C3*, or possibly a *Teensy*, with a *microSD* card for the archive. Power runs from a *5V solar panel* into a *LiFePO4* charger and four cells; without sun it should hold for up to three days. The light is *amber* and shines through the two **light-guiding antennae** *(PMMA rods, acrylic seeded with scattering particles, so a rod glows along its full length, not only at the tip).* Amber sheds the least of the blue that pulls in night insects, which counts for something in a valley that is meant to stay dark. Next to the light, the second channel is *sound*: a small speaker in the body plays a quiet sound field built from what the roots are hearing, and at a gathering it can be amplified live. Presence is watched by a *24 GHz radar*, the sensory roots plug in through individual connectors, and the printed parts are *PETG-CF* for strength. On top of that I am thinking of printing a few elements in photoluminescent *PETG (Ultraglow)*, which charges off daylight and then glows faintly on its own after dark, no power involved. The object should reach the field at the end of July, so I hope that next time I can write with fresh experience. The aim is to bring the unit to the point where it works on its own. **It charges from the sun, goes without maintenance over the long run, and speaks to the outside only when there is someone to hear it.** At the same time it can keep recording what it hears in the soil, building a local archive. In time I would like to grow the project into a network of several objects in the landscape that pass information between themselves, forming a shared technological (eco)system. ![[antennae-field.jpg]] --- ## [Second Brain Builder](https://kindl.work/Resources/Second+Brain+Builder) Update I rewrote the [building prompt](https://kindl.work/Resources/Second+Brain+Builder) for my **Obsidian Vault + LLM** system again. As you may know, Anthropic released the Fable 5 model, so I used it mainly to improve the whole system. Working with memory, context and data is a bit better again. Lately I have had good results working directly in [Max](https://cycling74.com/), [TouchDesigner](https://derivative.ca/), and even [Higgsfield](https://higgsfield.ai/) for visualisations, through MCP and an agent inside the Vault.