Next-Level LMS with Azure Custom Text-to-Speech Avatars | Learning Experience

About customer

The most in-demand format is a screencast, allowing real-time demonstrations of various technical tasks. However, maintaining the relevance of such videos presents a significant challenge. IT products evolve rapidly, and the constant need to re-record with a live trainer requires considerable time and effort, which is often practically unfeasible. Consequently, many videos become outdated and lose their relevance. Imagine watching a video that is not the freshest, or if you are a trainer, having to spend several days re-recording a two-hour lecture. This creates significant difficulties and hampers the learning process. In this case we describe the project which was started in order to improve our own LMS, which we use for teaching.

Awara IT is a global IT consulting company. Among its clients, there are both international and local companies worldwide. Offices are located in Kazakhstan, Spain, Italy, Finland, Bulgaria, and the USA. Awara IT automates and innovated key business processes digitally based on world-class technologies, including ERP, CRM, Low-Code, Data & AI, Business productivity, and Cybersecurity tools.

Starting point

Awara IT has been actively involved in the professional education market for many years, focusing on IT technologies, particularly those related to Microsoft products. In collaboration with numerous universities, our specialists train trainers, teachers, students, and administrators, while also developing the technical community. With the increasing volume of content, the availability of up-to-date, pre-recorded lectures on modern IT technologies has become crucial.

Challenges and solutions

Technologies

Azure OpenAI Service, Azure ML Studio, Azure Speech Studio, Azure neural voice, Azure Custom Text-to-Speech Avatar.

Implementation

Before using an avatar, courses were created on the Teachable platform using regular video recordings. After the introduction of the avatar, the process became significantly easier: now, to create a course, it is enough to write a text, and based on this text, a video is recorded using the avatar. Thus, human participation in video recording is no longer required.

We considered options such as animation with drawn characters and other solutions for creating digital avatars. In each case, aspects such as animation quality, realism, development and operational costs, and ease of integration with our existing platforms and processes were analyzed.

In figures:

6 content creators from our team.
300+ students
20k unique visitors’ coverage of video content on social networks.

Changing the recording of a training session with a live person takes 2-3 working days and costs $500 for studio rental. Making changes using a digital avatar takes less than 10 minutes and incurs no additional costs.

This solution offers numerous benefits:

For course administrators: Content updates are simpler and faster.
For trainers: They can focus on technical tasks, simply providing text for recording.
For the marketing department: The ability to create attractive course invitations and short videos using digital avatars.

There are also significant advantages for learners:

More personalization and engagement.
Quick access to up-to-date content.
Support for multiple languages, exceeding the capabilities of a single trainer.
Better sound quality than in non-professional recording studios.

Results

Before the implementation of the products, updating the content of training courses could only be done a few times a year. After the implementation, updates occur up to several times a day, including not only the correction of technical inaccuracies but also the improvement of material delivery methods based on user feedback.

The implementation caused following positive business impact:

Revenue costs: by reducing the cost of producing videos by at least 10 times, the gain amounted to approximately 90 thousand dollars per year.
Improvement adoption rates: The availability of a playground and the ability to work without writing code have enabled course administrators, who are not IT specialists or programmers, to independently create and record videos.
User satisfaction rates: 80% of users noted an improvement in speech quality. 76% of users expressed satisfaction with the avatar's representation compared to a live person.
Decreased IT spend: It is worth noting that it is no longer necessary to store raw course recordings for future editing. Additionally, there is no need to purchase expensive graphic packages for video editing. This results in significant savings on data storage, studio raw recordings, and the acquisition of costly editing software, which are no longer needed.