Virtual Assistant for Real Time Agricultural Insights – An example using Jasper, Alexa & Google AIY

Spread the love

On Day to day basis farmers needs weather insights as well as storage conditions at various storage units. They need real time insights on different agriculture related information. One of the options is for a tailored virtual assistants which can interact with users and internet sites on a regular basis giving insights. In this blog I tried to give an overview of different virtual assistants which are available both commercially and also open source level and how to develop/tailor one based on user needs.

windows_IoT

Virtual assistants have been used in many applications for day to day tasks. Virtual assitants helps users in automating tasks and also for getting real time insights and playing music. These Virtual Assistants work with minimum human interaction and whenever they communicate they use the natural voice of the users. The Assistants consists of 3 layers of architectures namely i. Speach to Text Engine ii. Logic Engine and iii. Text to Speech Engine. Speak to Text engine converts the user’s speak into text string so that it can be processed by the logic engine. The Logic Engine is the software component which receives text string from Speech to Text Engine and processeds the information and retrieves output to Text to Speech Engine. It uses a series of If Else statements or other API logic which can retrieve complext information and finally Text to Speech so that it can be more human centric.

The available commercial Virtual Assistants include i) Google Assistant ii) Cortana by Microsoft and iii. Siri by Apple. The commons Speech to Text which are avaialble to create any custom Virtual Assistants include Google STT, PocketSphinx, AT&T STT, Juluius, Wit.ai STT and IBM STT. The concepts around how to build a Virtual Assistant and its components are exmplained very well in a book by Tanay Pant. You may need to go through the deprecated code and some errors while installing pygame. The Virtual Asistant Mellissa can be developed ina  step by step activity or even it can be directly cloned to Raspberry Pi using Git. It has used PortAudio API along with PyAudio for binding with python, to develop the assistant.

Mellissa_Screenshot

Pic Source: Melissa

In this blog I would like to discuss the outcome of development of Virtual Assistant using Jasper and Google AIY. Jasper is an open source repository available via github. It has prebuilt images of the disk available for direct consumption. Alternatively it can also be used with manual step by step instructions mentioned in their Documentation site.

Google AIY Voice Kit is a latest add on in the tools for building projects using Raspberry Pi. This voice Kit comes with intelligent speaker, voice recognition and google assistant. Details about how to set up this system is given in AIY Projects site. This module can be used to prototype quickly using python based library using gPRC library and general bindings for languages like Go or Java. Another highly interesting use of pre built API was by TheRaspberryPiGuy on amazon Alexa personal assistant using Raspberry Pi and SenseHat. When installing Windows IoT on Raspberry Pi there is option of using Cortana.

Google_AIY_LED

Building Virtual Assistants in native languages to give a real time insights into weather, farmer tips and other market informaiton will not only help farmers in managing competitiveness but will also add value to the societal causes. The availability of technology and tools make development of such scenarios feasible but also can be developed with reasonble pricing factor. These type of tools along with systems which can monitor plant health in terms of NDVI value or soil moisture conditions giving real time feeeback systems to end users will help moving the agriculture to next level. If you want to play with these innovations checks the sites around Jasper, Google SDK for voice, Windows IoT, Amazon Alexa and other good books on those topics.

Thanks – Jak

Leave a Reply