Welcome everybody, we are back with news. Now we start again to do a weekly reports about our developments.
We launched and tested the English speech recognition module. The recognition quality is good. The problem of quickly adding new words for recognition has been solved. Now we can quickly and easily add rare words and toponyms, such as “appeltaart”, “singel” and others.
Work with open ontologies
The work with open ontologies has been completed. They have been converted into a structure that allows support programs to easily operate with knowledge. Also search, analysis and graphical representation of knowledge has been written for them.
With the help of this knowledge we will be able to significantly expand the picture of the world of AI, which is necessary for its development and further learning through reading texts.
Smart Mirror prototype
A prototype of the “Smart Mirror” is created - it is a device with the possibility of fully autonomous operation: it recognizes the voice and contains the core of artificial intelligence.
The device has demonstrated good performance. Smart mirror is not a product. It will be used as one of the test benches to demonstrate the capabilities of the core in various areas of human life: from individual smart devices to the global ecosystem around Artificial Intelligence.
Checked the work of the AI on the Raspberry Pi 3. As expected, it took some minor changes to accommodate some of the features of single-board computers.
As a result, we got a great response time of the AI, given the fact that the AI used only one computer core.
We also identified the modules that will need to be optimized for maximum performance in the future.
New memory layer
We have included another layer of memory. This time the seventh. It is responsible for the storage and the ability to output audio and images, that is, the original data.
We tested the interactive capabilities of AI. To do this, we added knowledge about insect bites, some knowledge about Amsterdam, and talked with the AI on these topics.
The test results indicated some points that can be optimized. The ability of the core to intercept the initiative and insert its own remarks on the topic of the dialogue, and, of course, to answer questions, was tested.
After the first development cycle was completed, extensive testing was conducted. The results were good and the technology once again proved its viability. Some flaws and omissions were also identified. All changes and additions were made to the next development cycle.