AI In A Box Envisions AI As A Private, Offline, Hackable Module

[Useful Sensors] aims to embed a variety of complementary AI tools into a small, private, self-contained module with no internet connection with AI in a Box. It can do live voice recognition and captioning, live translation, and natural language conversational interaction with a local large language model (LLM). Intriguingly, it’s specifically designed with features to make it hack-friendly, such as the ability to act as a voice keyboard by sending live transcribed audio as keystrokes over USB.

Based on the RockChip 3588S SoC, the unit aims to have an integrated speaker, display, and microphone.

Right now it’s wrapping up a pre-order phase, and aims to ship units around the end of January 2024. The project is based around the RockChip 3588S SoC and is open source (GitHub repository), but since it’s still in development, there’s not a whole lot visible in the repository yet. However, a key part of getting good performance is [Useful Sensors]’s own transformers library for the RockChip NPU (neural processing unit).

The ability to perform things like high quality local voice recognition and run locally-hosted LLMs like LLaMa have gotten a massive boost thanks to recent advances in machine learning, and it looks like this project aims to tie them together in a self-contained package.

Perhaps private digital assistants can become more useful when users can have the freedom to modify and integrate them as they see fit. Digital assistants hosted by the big tech companies are often frustrating, and others have observed that this is ultimately because they primarily exist to serve their makers more than they help users.

Continue reading AI In A Box Envisions AI As A Private, Offline, Hackable Module”

Bike Helmet Plays Music Via Tiny Motors For Bone Conduction

[Matlek] had an interesting problem. On one hand, a 40 minute bike commute without music is a dull event but in France it is illegal for any driver to wear headphones. What to do? Wanting neither to break the law nor accept the risk of blocking out surrounding sounds by wearing headphones anyway, and unwilling to create noise pollution for others with a speaker system, [Matlek] decided to improvise a custom attachment for a bike helmet that plays audio via bone conduction. We’ll admit that our first thought was a worrisome idea of sandwiching metal surface transducers between a helmet and one’s skull (and being one crash away from the helmet embedding said transducers…) but happily [Matlek]’s creation is nothing of the sort.

A 3D printed rack and pinon provides adjustability and stable contact with the “sweet spot” behind each ear.

The bone conduction is cleverly achieved by driving small DC motors with an audio signal through a TPA2012 based audio amplifier, which is powered by a single 18650 cell. By using motors in place of speakers, and using a 3D printed enclosure to hold the motors up to a sweet spot just behind the ears, it’s possible to play music that only the wearer can hear and does not block environmental sounds.

[Matlek] didn’t just throw this together, either. This design was the result of researching bone conduction audio, gathering a variety of different components to use as transducers, testing which performed best, and testing different locations on the body. Just behind the ear was the sweet spot, with the bony area having good accessibility to a helmet-mounted solution. Amusingly, due to the contact between the motors and the rest of the hardware, the helmet itself acts as a large (but weak) speaker and faint music is audible from close range. [Matlek] plans to isolate the motors from the rest of the assembly to prevent this.

Another good way to get audio to transmit via bone conduction? Send it through the teeth. While maybe not the best option for a bike rider, biting down on this metal rod sends audio straight to your inner ear.