SABOT | SABO Mobile IT

SABOT AI: In the bakery

SABOT is the ideal assistant for supermarket bakeries, enabling them to always offer fresh baked goods while conserving resources.

SABOT demo at EuroShop 2023

See SABOT in action in this short demo from EuroShop 2023. SABOT works smoothly even in a noisy environment.

SABOT Demo

Let's have a look at what SABOT can do!

BASIC INFORMATION

Speech - the basic skill of the human being

We learn our native language in preschool and use it throughout our lives to communicate with the world around us.

Machines are becoming more complex, their operation must be learned, and this requires attention. Advances in AI now make it possible to control and monitor machines in human language.

Thanks to the AI in SABOT, we no longer have to learn a set of commands and think about how to phrase a question so that it is understood by the machine. It is enough to express it in plain common language. If we are not understood with certainty, or if our wording is unclear, contradictory, or imprecise, SABOT does what a human listener would do: it asks questions until a clear, certain understanding is reached.

Spoken instructions or questions can also be given at a distance, allowing our eyes and hands to remain focused on the work at hand. This is especially useful when hands come into contact with food, need to remain germ-free while handling patients, or when using touchscreens with gloves is uncomfortable.

There are several thousand languages in the world, not to mention dialects. This makes no difference to the AI: any language can be learned.

TECHNICAL EXPLANATION

Language

Talking and listening without mouth and ears

The acoustic conditions in which you need to interact with the machine are as varied as the applications themselves. SABOT is a platform that allows multiple inputs and outputs of voice signals: from a built-in room microphone to soundproof headphones, from a clip-on microphone to input from a phone or smartwatch.

The conversion of the speech signal into text, and vice versa, the generation of the artificial voice is performed by online services that are tailored to the application, depending on requirements and cost. And who is listening? No one who isn't supposed to. Security is a top priority. SABOT's business model is not to sell data, but to provide machines with a Voice User Interface (VUI) in a professional environment.

AI Knows its Stuff

Machine learning for a lifetime

So that SABOT can enter into dialog, knowledge about the machine, the process it performs and the conditions in which it operates must be learned. Above all, safety is important: misunderstood commands should not cause any damage.

And as with humans, knowledge deepens with practice. Machine learning in SABOT is also based on this principle; each new situation is evaluated and added to the knowledge base. In this way, series products can be perfected during operation.

SABOT for any Machine

SABOT is a platform for retrofitting existing machines.

It includes tools for connecting to control systems and communication interfaces, to the machine, but also to IoT platforms, data archives, and IT that monitors and controls the machines, as well as coupling to sensors outside the machines.

Learning and testing dialogs can be simulated faster and safer in SABOT without the physical machine.

Coupling with the existing graphical user interface provides smooth and consistent operation. This allows you to provide visual feedback or display graphics that are easier to understand visually than long explanations. SABOT is not a plug-and-play product. However, it does allow for safe and efficient customization.

Quick Integration

Quickly achieve demonstrable benefits

The first three questions are: Is voice control possible and practical for a particular machine, and will users accept it? You can try it with little effort. Now that you have an idea, let's talk about it!

How it all works

The core of the communication with the operator is the speech processing. It converts the spoken language into text using the STT component. The text is further processed by a language model, whether based on NLP or, more recently, LLM models, which recognizes the user's intentions.

Based on the context and previous conversations, the system either generates a response or triggers a recognized action. The executed action usually leads to a result, which is processed repeatedly by the language model, which finally generates a response for the user. The system also responds to external stimuli, primarily from the machine itself, but input from other sources such as the business intelligence system and the like can also be taken into account.

The operator then receives responses in the form of speech converted from text using TTS components.

MORE INFORMATION

Does SABOT understand our specific vocabulary?

Newer LLMs are trained on large data sets and are able to recognize most human words, even technical words. If it is something very specific, the models can be refined with additional terms.

Is it secure? Is it listening to my private or customer conversation?

There are several ways to ensure that only the expected speech is processed. SABOT only starts processing speech after the activation word has been recognized. Activation words are usually processed locally and after positive recognition the speech is processed by STT. Speaker recognition can be adapted to the respective operational requirements so that the usage time, for example, must match the working hours, etc.

How is SABOT integrated into our machine?

SABOT was designed as an open system. It can be operated as a complete stand-alone solution or easily integrated into the machine system thanks to its multi-service architecture. On the SABOT HW hardware, a set of SABOT services are available to manage the primary audio inputs and outputs. Most SABOT services run on a backend server and perform more complex operations.

What are the costs related to SABOT?

Generally, it is the number of operator inputs processed and the number of machines managed in the system. The most cost-intensive services are provided by cloud providers. It is mainly the conversion of speech to text and the subsequent recognition of meaning. There is also the option of using local solutions for STT or NLP, depending on the customer's requirements.

GET IN TOUCH

Thank you! Your submission has been received!

Oops! Something went wrong while submitting the form.

AI-driven voice communicationwith machines

AI-driven voice communication
with machines