Technology

Abstract representation of a network with connections.

Distributed processing

Every single nnounce platform device has it’s own dedicated & powerful DSP. In combination with centralized management of DSP design templates a fully distributed processing system can be built. This approach alleviates many problems which plague centralized processing platforms. Most importantly it greatly simplifies network requirements and improves network reliability since there are no central choke & failure points in the system. All network traffic flows directly from input device to output device and takes the shortest path possible.

The platform comes with a web-based free-form audio processing pipeline design studio with rich audio processing component library built into every device. All devices share a common set of design tools and components.

Software defined architecture

At the heart of every nnounce device sits a heterogenous multi-core processor. The processor cores are specialized - general processing cores, DSP core, AI accelerator and a real-time core. Every core is utilized for what it is at best at - tasks are distributed and isolated according to product configuration.

Specialized cores and task distribution allow nnounce devices to be configured for different applications - from higher network channel processing counts to user interface oriented products.

The nnounce platform products utilize just a fraction of capabilities of the underlying hardware - extensive capacity for product evolution is built into them.

Close-up of a device with multiple input jacks and cables plugged into them.

Audio distribution and interoperability

The platform’s low latency network stack begins with utilizing built-in advanced programmable ethernet switch functions to sort between latency sensitive audio traffic and best effort traffic. Pre-sorted traffic is entering the CPU trough completely separated pipelines utilizing multiple hardware queues and using expedited linux™ network processing path AF-XDP. From there the audio flows straight into DSP for processing or sending to audio outputs.

This architecture enables nnounce devices to spend significantly less time during network-to-dsp and dsp-to-network transfer. Thanks to this and the distributed point-to-point nature of the system, more time can be allocated to actual time spent on network, leading to greatly improved network reliability.

Close-up of the olive colored mesh against the bardo color background.

Integration tools

The platform’s integration toolbox offers many options for advanced integration. All devices are equipped with either dry contact inputs & outputs or GPIO. In the spirit of distributed intelligence, scripts and applications can be executed in individual devices while still being managed centrally in main configuration application. The devices can also be controlled remotely through basic REST standard communication API’s.

For advanced applications, the nnounce platform enables users to deploy and run a full application container developed in their chosen language using the APIs provided by the nnounce platform.

On a black background, blue-colored electricity bursts out of the sphere.

Artificial intelligence

The use of artificial intelligence based on neural networks is growing rapidly in all industries. It has revolutionized many applications and pushed many applications to a completely new level of quality. The nnounce platform devices contain a neural network accelerator that can be plugged into the audio processing pipeline. This design enables novel applications, such as neural network based background noise removal, ambient noise compensation, audio codecs, gunshot detection, speech-to-text and text-to-speech conversion. With the embedded neural-network accelerator, nnounce platform devices are ready for any future application.

A laptop with green code on the screen in the style of "The Matrix".

Security

The nnounce platform & devices offer unparalleled hardware, firmware and software stack security. Security has been a key requirement from the beginning of the platform’s inception. The entire platform and all devices - hardware, firmware and software - share a common security stack. The device’s firmware & software stack is built on the rock-solid foundations of the Rust programming language. All core software is kept as close as possible to the mainline software releases. The amount of proprietary patches is minimized. Thanks to this it is easy to keep the entire security & OS stack always up to date.

The nnounce platform also offers extra privacy for sensitive applications by offering an optional AES67 encryption.