JIMMA INSTITUTE OF TECHNOLOGY

FACULTY OF COMPUTING AND INFORMATICS

INFORMATION TECHNOLOGY PROGRAM
FINAL YEAR PROJECT-1
“VOICE TRANSLATOR APP (AMHARIC, …) ”

Group member's name ID no.

1. Yirgalem Dereje Eshetu RU 4653/14
2. Habtamu Takele Bidika RU 0553/14
3. Halid Abdurehman Ali RU 0272/14
4. Gutera Getahun Joba RU 0919/14

A Final Year Project proposal

Submitted to Faculty of Computing and Informatics, Jimma Institute of Technology,
Jimma University, in Partial fulfillment of the requirement of the Degree of Bachelor
of Science in Information Technology

Jimma, Ethiopia
November 2023/24

i
 Table of Content
Abstract ........................................................................................................................ 1
1. Introduction .............................................................................................................. 2
2. Statement of the Problem ......................................................................................... 3
3. Objectives of the Study ............................................................................................ 3
3.1 General Objectives .......................................................................................... 3
3.2 Specific Objectives ..........................................................................................3
4. Proposed System ...................................................................................................... 4
5. Scope of the Project ................................................................................................. 6
Core Functionalities: ............................................................................................. 6
Technical Specifications: .......................................................................................6
Limitations: ............................................................................................................6
6. Methodology ............................................................................................................ 7
7. Development Tools .................................................................................................. 8
Software Tools .......................................................................................................8
Hardware Tools ..................................................................................................... 8
8. Limitations of the Project .........................................................................................9
9. Budget .................................................................................................................... 10
Required Resources and Costs ............................................................................ 10
Total Costs: ..........................................................................................................11
10. Project Timeline and Milestones ..........................................................................11
11. Team composition ................................................................................................12

List of Tables
Table 1 : software development tools ....................................................................8
Table 2 : Hardware development tools ................................................................. 8
Table 3 : Cost estimation of the project .............................................................. 10
Table 4 : Timeline and Milestone ........................................................................12
Table 5 : Team composition ................................................................................ 12

ii
 Abstract

This project develops a Voice Translator App to break language barriers through
real-time speech translation. It captures spoken language, translates it instantly, and
outputs the result in the target language. Using speech recognition, machine
translation, and text-to-speech technologies, the app supports multiple languages
with an intuitive interface. By leveraging TensorFlow Lite and PyTorch, the app is
optimized for mobile devices and works offline by storing essential data locally.

Key features include accurate translations, real-time performance, and easy usability.
The app is designed for seamless communication across cultures, offering a practical
solution for travelers, business professionals, and anyone in need of quick translation.

1
1. Introduction

1.1 Background of the Project

Project Selection
In today’s world, many people speak different languages, and this can cause
problems when trying to talk to each other. This project aims to solve that problem
by making a Voice Translator app. The main idea is to create an app that can
translate languages in real time, so people who speak different languages can talk
easily and understand each other.

Project Goals

 Real-time Translation: Make a system that can change spoken language into
another language while talking.
 Multilingual Support: Include many languages to help users from different
countries.
 User-Friendly Interface: Create an app that is easy for everyone to use, even
if they don’t have much experience with technology.
 Offline Functionality: Make the app work without the internet, so people in
places with no internet can still use it.

Interest in the Topic

The world is becoming more connected, and people need to communicate with
others who speak different languages. A voice translator app can help in many
situations, like business, travel, and helping people in need. This project can make it
easier for people from different countries to work together and understand each other.

Prior Work
Many translation tools already exist, but most of them only work with text or need a
lot of effort from the user. This project will bring new technology to language
translation, making it easier for people to talk to each other no matter where they are
from.

2
2. Statement of the Problem

Language barriers are still a big problem when people try to communicate and work
together in today’s connected world. Traditional ways of translating, like using
dictionaries or asking a human translator, can take a lot of time, cost money, and
don’t always work well for all situations.

The main problem that this project will solve is the lack of a fast, reliable, and
accurate way to translate spoken language in real time. Right now, most translation
tools only work with text and might not give correct translations for longer sentences
or more complicated meanings.

Because of this, there is a clear need for a voice translator app that can help people
communicate easily, even if they speak different languages. The app should provide
accurate translations and help break down language barriers in conversations.

3. Objectives of the Study

3.1 General Objectives

1. To create a strong and easy-to-use voice translator app that can translate
languages accurately and in real time, while keeping the meaning of what is
said.
2. To help people from different cultures communicate and work together more
easily by providing a smooth and effective language translation solution.

3.2 Specific Objectives

1. To develop a speech recognition system that can clearly listen to and turn
spoken words into text, even in noisy environments.
2. To connect the app with a reliable translation service (like Google Translate
or another API) to make sure translations are accurate and meaningful.
3. To build a text-to-speech (TTS) system that can read the translated text out
loud in a clear, natural-sounding voice.

3
 4. To design a simple, easy-to-use app interface that anyone can navigate,
whether they are tech-savvy or not.
5. To make the app run smoothly on different devices, like smartphones, tablets,
and even older models, by improving its speed and efficiency.
6. To test the app’s performance, making sure it works well in real-world
situations, and gather feedback from users to make it better.
7. To ensure the app can work offline, allowing users to translate even when
they don’t have an internet connection.
8. To support multiple languages, covering the most common languages spoken
worldwide, to help a large number of people.
9. To include voice recognition that can understand different accents and
pronunciations, making the app more accurate in various regions.
10. To provide a security feature that keeps user data and translation history
private and safe

4. Proposed System

The proposed Voice Translator app will help people communicate easily with others
who speak different languages. The app will use modern technology to translate
spoken language into another language in real time. It will work by combining
speech recognition, machine translation, and text-to-speech systems.

Key Components:

 Speech Recognition Module:

 Uses a strong speech recognition system to listen to and turn spoken words into
text.
 Supports many languages and different accents to make the system more
accurate.
 Includes features like noise reduction and echo cancellation to make sure the
audio quality is clear, even in noisy places.

 Machine Translation Module:

 Connects to a translation service (like Google Translate or similar) to change the

spoken text into another language.

4
 Uses advanced translation methods (like neural machine translation) to give
better, more accurate translations.
 Supports many different language pairs to help users from all around the world.

 Text-to-Speech Module:

 Turns the translated text into natural-sounding speech using a high-quality text-
to-speech system.
 Allows users to choose different voices and accents to match their preferences.
 Makes sure the speech is clear, even in noisy environments, so users can easily
hear the translation.

 User Interface (UI):

 Designed to be simple, easy to use, and intuitive for all kinds of users.
 Supports both text and voice input and output. Users can type or speak to
communicate.
 Provides real-time feedback to users, so they can see and hear translations
instantly.
 Includes offline capabilities, so the app can still work without internet access in
places with limited connectivity.

By combining these components and leveraging advanced technologies, the

proposed Voice Translator application aims to revolutionize cross-cultural
communication

5
5. Scope of the Project

The goal of this project is to develop a strong, easy-to-use voice translator app that
can translate spoken language accurately and quickly in real time. The app will focus
on the following main features:

Core Functionalities:

 Speech Recognition: The app will listen to and convert spoken words into
text with high accuracy.
 Machine Translation: It will translate the text into another language, making
sure the translation is clear and meaningful.
 Text-to-Speech: The app will turn the translated text back into speech,
making sure it sounds natural.
 User Interface: The app will have a simple, easy-to-navigate interface so
users can use it without trouble.

Technical Specifications:

 Supported Languages: The app will support many common languages used
around the world.
 Offline Capability: It will have limited offline features for basic translation
when there is no internet.
 Platform Compatibility: The app will work on smartphones and tablets, so
people can use it on different devices.
 Performance: The app will provide real-time translations with minimal delay.
 Accuracy: The app will focus on being highly accurate, especially for
common phrases and sentences.

Limitations:

 Dialect and Accent Variations: The app may have trouble understanding
strong accents or regional dialects.
 Complex Language Nuances: The app may not always handle complex
phrases, idioms, or very detailed meanings perfectly.

6
  Internet Connectivity: The offline version will have limited accuracy
compared to when the app is connected to the internet.

This project will mainly focus on building and testing the key features of the voice
translator app. In the future, additional features like language detection, accent
recognition, and better offline capabilities could be added to improve the app.

6. Methodology

1. Project Planning: Define the app's core features (speech recognition, machine
translation, text-to-speech) and performance goals (accuracy, speed, offline support).
2. Technology Selection: Choose the platform (Android, iOS, or cross-platform) and
development tools (e.g., Android Studio, Xcode). Use AI frameworks like
TensorFlow Lite or PyTorch Mobile for model deployment.
3. Data Collection: Gather and preprocess speech, translation, and text-to-speech
data, ensuring it's clean and augmented for accuracy.
4. Model Training: Train core models (speech-to-text, translation, text-to-speech)
and optimize them for mobile performance.
5. App Development: Build the app with an intuitive UI, integrating speech
input/output, offline support, and error handling.
6. Testing: Conduct unit, integration, and user tests to ensure functionality, speed,
accuracy, and usability.
7. Deployment: Release the app on Google Play Store and Apple App Store,
providing regular updates and user support.

7
7. Development Tools

Software Tools
Purpose Tools/Programs
Operating system for development and Android (Android Studio), iOS (Xcode)
testing.
For real-time speech recognition Google Speech-to-Text API, IBM
(convert speech to text). Watson Speech to Text
For translating text into different Google Translate API, Microsoft
languages. Translator API
For converting translated text into Google Text-to-Speech API, Amazon
natural speech. Polly
For designing user interface and creating Figma, Adobe XD
mockups.
For browser-based testing and Google Chrome, Firefox
debugging.
For documentation, reporting, and MS Word, MS Excel
project tracking.

Table 1: software development tools

Hardware Tools

Purpose Tools/Programs
For storing development data and project Hard Disk (64 GB)
files.
For transferring and testing the app builds. Flash Storage (64 GB)
For coding, testing, and managing Laptop/PC with at least 16 GB RAM
development.
For testing speech recognition Microphone (preferably noise-
(microphone). canceling)

Table 2: Hardware development tools

8
8. Limitations of the Project

While the proposed voice translator application aims to bridge language barriers,
there are certain limitations to consider:

1. Language Support: The application's effectiveness is dependent on the

quality and availability of language models and translation APIs. Limitations
in language support, particularly for less common or regional languages, may
impact the accuracy and fluency of translations.
2. Dialect and Accent Variation: The application may struggle to accurately
recognize and translate dialects and accents, particularly those that deviate
significantly from standard pronunciations.
3. Complex Linguistic Nuances: Highly complex linguistic nuances, idiomatic
expressions, and cultural references can be challenging to translate accurately.
4. Real-time Performance: Real-time translation may be affected by factors
such as network connectivity, device processing power, and the complexity
of the input language.
5. Offline Functionality: Offline capabilities will be limited, as they rely on pre-
trained models and data. The accuracy of offline translations may be lower
compared to online translations.
6. Ethical Considerations: The use of language technology raises ethical
concerns, such as bias in language models and the potential misuse of
translation tools.

9
9. Budget

Required Resources and Costs

Category Item Description Estimated Cost Estimated Cost
(ETB) (USD)
Hardware Personal computer (for 47,000 $381.70
development and training
models)
Mobile device (for testing and 19,000 $154.22
debugging)
Software Development tools (Android Free Free
Studio, TensorFlow Lite, etc.)
Other necessary software Free Free
(e.g., Git, Python)
Reference Textbooks and online courses 4,000 $32.46
Materials (machine learning, NLP)
Online research resources and 2,000 $16.23
subscriptions (optional)
Internet Access Data for research, 5,000 $40.57
development, and training
Deployment Google Play Store (one-time 3,080 $25
developer account
registration)
Apple App Store (annual 12,217 $99
developer account fee)
Hosting Costs (for backend 7,000 Approximately
services/APIs) fixed
University Use of university lab Provided by Provided by
Resources computers and internet university university
Miscellaneous Contingency fund for 3,000 $24.34
unexpected expenses
Table 3: Cost estimation of the project

10
Total Costs:

The total estimated cost for the Voice Translator app project, based on hardware,
software, internet access, and deployment, is approximately 99,297 ETB or $773.52
USD.we need to use the current exchange rate for USDT to Ethiopian Birr (ETB).
As of now, 1 USDT is approximately 123.19 ETB.

10. Project Timeline and Milestones

Month Week Tasks/Milestones
Month 1: Planning Week 1 Define project scope, identify target languages, and
and Setup set core features (speech recognition, translation,
TTS).
Week 2 Select platform (Android/iOS), development tools
(Android Studio, Xcode), and machine learning
frameworks (TensorFlow Lite).
Week 3 Set performance goals (accuracy, speed, offline
support) and define success criteria.
Week 4 Set up version control (GitHub) and research
translation APIs.
Month 2: Data Week 5 Gather and preprocess datasets for speech-to-text,
Collection & translation, and TTS.
Model Training
Week 6 Apply data augmentation techniques and format
datasets for training.
Week 7 Train speech-to-text and machine translation
models.
Week 8 Continue training models and begin work on text-
to-speech (TTS).
Month 3: App Week 9 Design UI/UX for simple navigation and user
Development & interaction.
Feature
Integration
Week 10 Integrate speech recognition and TTS functionality.

11
 Week 11 Implement machine translation and offline
capabilities.
Week 12 Optimize AI models for mobile and integrate error
handling.
Month 4: Testing, Week 13 Conduct unit and integration tests; start alpha
Evaluation & testing with users.
Deployment
Week 14 Test for edge cases, accents, and offline
functionality.
Week 15 Evaluate performance (speed, accuracy, battery)
and refine features.
Week 16 Finalize app for deployment, submit to app stores,
and prepare for maintenance.

Table 4: Timeline and Milestone

11. Team composition

S.N Full Name Id. No: Email/Mobile Responsibility

o.
1 Gutera Getahun RU [Your Email] Project Lead: Manages timelines,
Joba 0919/14 documentation, app design, and speech-
to-text.
2 Habtamu RU [Your Email] Backend Developer: Develops server-
Takele Bidika 0553/14 side architecture, APIs, and integrates
machine translation.
3 Halid RU [Your Email] Frontend/UI Developer: Designs the
Abdurehman 0272/14 UI and integrates speech recognition and
Ali TTS features.
4 Yirgalem RU [Your Email] QA and Testing: Ensures app
Dereje Eshetu 4653/14 functionality, performance, and verifies
speech accuracy.

Table 5: Team composition

12