Client Case Studies
Real scenarios. Real outcomes. Real reliability.
Choose industry:
CASE STUDY
Text To Audio -Universal IM Platform
RESULTS
Multi-channel bot with API functionality.
Operating on a budget of under EUR 2/mo.
18 languages supported, with audio generation per language.
CONTEXT
A non-profit text-to-audio bot operating across five messenger platforms
The Text To Audio bot was built to support visually impaired users by converting any text message into spoken audio - directly inside the instant messaging apps people already use every day.
Challenge
Numerous text-to-audio converters exist, such as those from Google, IBM, and others, but they present several challenges
Most converters offer software development kits (SDKs) but lack accessible client applications for regular users.
Users often struggle to locate and install functional converter client applications.
Installing additional software becomes unnecessary when users can simply send a message to a bot - a more straightforward approach.
Application Infrastructure Requirements:
- A simple and user-friendly UI with a streamlined registration process
- Multiple voice options for audio conversion
- A straightforward admin tool for reviewing user-to-bot conversations to facilitate bot improvements
- Cost-effectiveness
HOW WE HELPED SOLVE THE PROBLEM
Serverless application comprising two distinct components:
- API with Viber, Telegram, WhatsApp, and Facebook Messenger bot integration
- An admin console website that enables authorized administrators to access and review bot usage logs
In developing Text to Audio, we extensively utilized AWS services (particularly DynamoDB and Lambda) to create a highly cost-effective application.
Technical Details
In building Text to Audio, we extensively utilized serverless AWS services - particularly DynamoDB, Lambda, and S3 - to ensure cost-effectiveness.
The Text To Audio bot is deployed on AWS Lambda, utilizing an S3 Bucket to store users' audio files and DynamoDB to log all user messages. It is accessible on Instagram, Telegram, Viber, WhatsApp, and Facebook Messenger. Alternatively, the bot can be deployed on a regular server with PostgreSQL without requiring re-compilation.
The admin console is a website designed for authorized administrators to review bot usage logs. The frontend is stored on an S3 bucket, developed using VueJS, and employs Cognito for authentication. The backend is deployed on AWS Lambda.
- Users send a message via Instagram, Messenger, WhatsApp, Viber, Telegram, or REST API.
- Bot forwards the request to API Gateway.
- API Gateway invokes a Lambda function to process the message.
- Lambda detects the language using an external API.
- Text is sent to IBM or Google Text-to-Speech API for audio generation.
- Generated audio files are stored in S3.
- Requests and responses are logged in DynamoDB.
- Admin opens the console in browser.
- DNS routing is resolved via Route 53.
- CloudFront serves the static site hosted on S3.
- User authenticates via Cognito.
- Console invokes Lambda via API Gateway to fetch messages.
- Lambda retrieves data from DynamoDB and returns JSON.
RESULT
Multi-IM channel bot with API functionality, operating on a budget of under $3/mo.
Try it yourself
Try the Text to Audio conversion now!
I so much like this your website and what you are doing for the visually impaired community like us in coming up with an idea to convert text to audio.
I sincerely love what you guys are doing please keep it up.
I am doing this as a volunteer service for persons with visual impairment like me converting their textbooks soft copy materials etc to mp3 audio.
The university and I also benefit students there and elsewhere across Nigeria.
It will be very easy and accessible for visually impaired users like me.
This good gesture by you.
Thanks and God bless.
From Uzo.
Nnamdi Azikiwe University Awka
Nigeria
Helping Africa
Nigeria