Speech Recognition Tool for an Offline Educational PWA

Home
Case Studies
Speech Recognition Tool for an Offline Educational PWA

Customer location

Project Duration:

1.5 years

Client is a US-based educational software provider focused on building free of charge solutions to foster global literacy. Being halfway through their signature product development, the client got stuck with the lack of engineering resources for building a new sophisticated computational tool. The expertise in question was deep learning, ETL, QA, and cross-platform development. Considering Symfa partnership approach, along with a track record in big data, machine learning and AI, and business automation, our team was selected as the closest fit for the client’s team extension.

The critical requirement was to ensure the app’s standalone functioning so as the internet connection breaches couldn’t be a blocker. To manage to introduce fault-tolerant speech recognition capabilities within any conditions, the team encoded the required techniques directly into the browser. As it was necessary to do without Amazon, Azure or Chrome speech APIs, we utilized a WASM module to deploy the native code to the browser while handling Javascript-enabled integration. To ultimately improve the app’s search performance, Symfa applied the database sharding approach. Also, the team made it to max out the browser’s recognition accuracy through extensive organic training and dedicated word stock implementation. For institutional-grade approval, the app content was fully aligned with the Common Core State Standards (CCSS).

Solution

Designed for targeted instruction and audiovisual support automation within the ELA classroom, the solution is a progressive web app able to smoothly function, regardless of the internet connection quality. The system flawlessly runs across the major browsers, enabling teachers to easily manage the studying process and organize an intuitively engaging experience for students. By providing a multitude of in-built educational challenges throughout the daily lexile reading practice, the app motivates students for greater learning achievements. Above that, the solution helps develop mental and verbal skills while aiding teachers and parents in assessing each child’s potential and progress.

Personalized libraries

Progress monitoring & stimulation

Student performance dashboard

Work done

Personalized libraries

Based on reading skills level, the solution automatically builds a dedicated collection of books for each student.

Unlimited texts for grades from K-12
Achievement-based selection updates

Numerous topic categories
Subject-matter articles and commentaries

Progress monitoring & stimulation

All the texts are adapted to naturally cultivate the agility of mind while continuously challenging student’s attention and measuring skills level.

Initial adaptive assessment
Embedded assessments within each text

Versatile lexile metrics implementation
Written-response, multiple-choice, and cloze tests

Student performance dashboard

The module gives educators real-time access to the stats on individual students and overall class comprehension, reading behavior, and knowledge growth over time.

Instructions and categories correction
Lexile scores monitoring
Achievement leaderboard

Individual reading goals setting and completion tracking
In-app social network for student-teacher communication

Work done

Existing functionality performance optimization
Comprehensive QA audit and bug fixing
Angular 5 to Angular 8 version update

Text and image compression functionality engineering
Speech recognition techniques implementation

Tech guidelines development for the client’s team
Testing documentation for the QA team

Angular
ASP.NET Core
Bootstrap
Node.js
SQL Server
PostgreSQL
DB Sharding
VS Code
Reshaper

Available for free, the solution quickly engaged a wide user audience while promoting literacy, garnering positive feedback, and building a tight-knit education contributors community worldwide. Symfa made a monumental investment to their teamwork karma by brilliantly tackling parallel product engineering and testing, irrespective of acting within distributed teams in need of non-stop cooperation.