Swiming-pool One Way "walkie-talkie" (or Rather "Swimie-Talkie")

The idea behind the project is simple: capture sound outside the pool, process it electronically, and reproduce it underwater through an underwater acoustic transducer. The result is a "transparency mode" that allows a swimmer to perceive sounds from the outside environment while remaining fully immersed.

This project is quite demanding in terms of tooling and it could not have happened without the help of JUSTWAY that are sponsoring this project by supplying high quality, quick turn around parts in both Aluminum (CNC) and Nylon (SLS 3D printed). I have now quite a bit of experience with JUSTWAY services with metal 3D printing, Nylon 3D printing, multi-color 3D printing and transparent resin 3D printing and I still am impressed with both the part quality as well as the customer service

This project is not intended to be a finished product. Instead, it is a prototype platform designed to validate the concept and provide a flexible starting point for experimentation. Many technical questions remain open, including the best underwater transducer design, the required acoustic power and the signal processing needed to maximize speech intelligibility. The purpose of this prototype is to answer these questions experimentally.

For this reason, the system is built around a Raspberry Pi making for an extremely flexible platform and a standard 18 V (BOSCH) cordless power tool battery

Rather than presenting a definitive solution, this project aims to provide an open and modifiable platform. Every aspect of the design from the electronics and mechanical construction to the software and acoustic components can be adapted, improved, or completely redesigned. Whether your goal is to build a more powerful underwater speaker, experiment with different diaphragm materials or develop advanced speech enhancement algorithms this prototype is intended to serve as a foundation for further exploration.

Soldering iron

Digital multimeter

Small screwdriver (for DC-DC adjustment)

Crimping pliers (optional)

Raspberry Pi (with rasbian)

18 V Bosch Professional battery

DC-DC Buck Converter

Dayton Audio DAEX58FP exciter

3.5 mm audio cable

Wires

2 part silicone (for custom seals making)

Access to a 3D printer and CNC machine. I used JUSTWAY services to make these parts

All the PCB design files as well as the CNC and 3D printed parts are available here: https://github.com/Virgile-Colrat/Swiming_pool_speaker

One of the first challenges encountered during this project was determining how to efficiently generate sound underwater. Unfortunately, underwater acoustics are very different from acoustics in air, and a traditional loudspeaker is poorly suited to this application. Indeed, water is approximately 800 times denser than air, meaning that moving water requires significantly more force than moving the same volume of air. As a result, the traditional audio speaker's efficiency drops dramatically and, in many cases, the driver may even be damaged because it is operating far outside its intended conditions.

Another important consideration is acoustic impedance. The acoustic impedance of water is much higher than that of air, making the transfer of acoustic energy between a conventional loudspeaker and the surrounding water extremely inefficient

For these reasons, the objective is not to place a loudspeaker underwater, but rather to create a vibrating surface capable of efficiently coupling its motion to the surrounding water.

Instead of using a conventional loudspeaker, this prototype uses an electrodynamic exciter. It consists of a moving voice coil attached to a rigid mounting surface. When driven by an audio amplifier, the exciter causes the entire surface to vibrate, effectively transforming it into a loudspeaker diaphragm.

Advantages:

1. The vibrating surface can be designed specifically for underwater operation.
2. The exciter itself remains protected inside the waterproof enclosure.
3. Different diaphragm materials and geometries can easily be tested without changing the electronics.
4. The mechanical properties of the diaphragm can be optimized independently of the actuator.

This last point is particularly important because, as will be shown later in the project, the diaphragm is arguably the most critical component of the entire underwater speaker. Its diameter, stiffness, mass, material, clamping method and resonance frequency all have a much greater influence on the final acoustic performance than the exciter itself.

Since the purpose of this project is to validate the concept rather than maximize acoustic output, the exciter was chosen according to several practical criteria (cost, availability, etc). After comparing several commercially available solutions, the Dayton Audio DAEX58FP was selected.

It is important to emphasize that this choice was not made because it is the optimal underwater transducer. Rather, it provides a practical and inexpensive actuator capable of driving a variety of experimental diaphragms while remaining easy to integrate into a waterproof enclosure.

The amplifier was selected using a similar philosophy. At this stage, the objective is simply to provide enough electrical power to drive the exciter while maintaining good efficiency and low power consumption from the battery.

A compact Class-D amplifier based on the TPA3110 integrated circuit was chosen. Class-D amplifiers typically achieve efficiencies greater than 85%, making them particularly well suited for battery-powered applications. Their low heat dissipation also simplifies

The Raspberry Pi requires a stable 5 V supply to operate correctly, while the chosen Bosch battery provides a voltage that varies between approximately 15 V (nearly discharged) and 21 V (fully charged). A DC-DC buck converter is therefore required to reduce the battery voltage to the regulated 5 V required by the Raspberry Pi.

Before connecting the converter to the Raspberry Pi, its output voltage must be adjusted. Most inexpensive buck converter modules are shipped with an adjustable output voltage and there is no guarantee that the factory setting is close to 5 V. Connecting an unadjusted module directly to the Raspberry Pi may permanently damage the board.

Power the converter from the battery and measure its output voltage using the multimeter (see picture). Using the small screwdriver, slowly rotate the adjustment potentiometer while continuously monitoring the output voltage.

Adjust the converter until the output reads: 5V

Most converters require several turns of the potentiometer before a noticeable voltage change occurs, so make small adjustments and monitor the multimeter carefully.

Once the output voltage has been set, disconnect the battery and reconnect it once more to verify that the converter consistently starts at approximately 5.0 V.

The second picture shows the assembly of all the electronic components. The amplifier itself can operate from a much wider voltage range than the raspberry pi and is directly powered by the 18V of the battery. The audio amplifier is connected to the raspberry pi through the jack connector. The amplifier is capable to output stereo signal but in this prototype, only one speaker is connected.

The enclosure must satisfy two seemingly contradictory requirements. It must remain completely watertight to protect the electronics, while simultaneously allowing the vibrating diaphragm to move freely. Achieving both objectives at the same time is one of the main challenges of the design.

The vibrating plate is clamped between two silicone seals positioned around its outer perimeter. This arrangement serves several important purposes simultaneously. First, the silicone seals provide the waterproof barrier between the inside of the enclosure and the surrounding pool water. No water should be able to reach the electronics located behind the diaphragm.

Secondly, because silicone is elastic, the diaphragm is not rigidly constrained by the aluminum clamps. Instead, it is allowed to move by a small amount in response to the force generated by the electrodynamic exciter.

This small freedom of movement is essential. If the diaphragm were directly bolted to the aluminum housing, its vibration would be severely restricted and much of the amplifier's energy would simply be transmitted into the enclosure instead of into the water. The diaphragm can therefore be viewed as "floating" inside the enclosure rather than being rigidly fixed to it. Perhaps the greatest advantage of this design is its modularity, because the diaphragm is simply clamped between two sealing rings, it can be replaced within minutes. Different materials, thicknesses, diameters and geometries can therefore be evaluated without modifying the remainder of the system. This is also true for the seals the hardness and thickness of which can be customized. This makes the enclosure not only an underwater loudspeaker but also an experimental platform for investigating how diaphragm design influences underwater acoustic performance.

To validate the concept as quickly as possible, the prototype uses the open-source Mumble voice communication software.

Both the Mumble server and a Mumble client are installed directly on the Raspberry Pi. The server receives audio transmitted over the local Wi-Fi network from a smartphone, while the client automatically connects to this local server and outputs the received audio through the Raspberry Pi's headphone jack, which is connected to the audio amplifier.

The primary reason for choosing Mumble is its immediate availability. Rather than developing a custom communication protocol or smartphone application, Mumble provides a complete, robust, and low-latency voice transmission solution that is freely available for Linux, Windows, macOS, Android and iOS. This allows the hardware platform to be validated without writing any custom networking software.

It is important to note that Mumble is used solely as a proof-of-concept tool. The objective of this project is to evaluate the underwater acoustic system rather than the communication protocol. Once the concept has been validated, the Raspberry Pi software can easily be replaced by a custom application tailored to the final user interface and desired features. For this reason, the Raspberry Pi was selected as the processing platform, providing the flexibility to evolve the software without requiring any hardware modifications.

This project worked as expected but the small size of the components make the speaker quite weak. Although the sound can be clearly heard in the water, the volume is quite weak so much so that with the current components, it hard to understand speech if you are over 1.5m away from the speaker. A new version would have a much bigger speaker with high power electronics but this version already serves as a good proof of concept.

It is important to understand that the components selected for this prototype are intentionally modest. The amplifier, the exciter, the underwater diaphragm, and the power system have been chosen because they are inexpensive, easy to obtain, and sufficiently powerful to validate the concept. They are not sized to provide high sound pressure levels throughout a large swimming pool. Their role is to demonstrate that the principle works and to provide a development platform on which improved transducers, more powerful amplifiers, better digital signal processing, or entirely different mechanical designs can later be tested.

One thing I will change in a future version is to have the ability to turn ON/OFF the electronics from the outside (probably with a simple magnetic switch).

Note that initially the custom PCB interfacing the battery with the raspberry pi had the 18V->5V converter on it but I fried it during the different testing steps and I had to use the “external” DC-DC converter I had laying around.