Abstract: Text-to-Speech (TTS) systems for low-resource languages like Igbo face significant challenges due to dialectal diversity. This research presents a dual-language TTS system for English and Igbo, specifically designed to harmonize the diverse pronunciations and linguistic features across different Igbo dialects. Leveraging a custom model trained on a curated dataset, the system aims to generate natural-sounding speech for both languages. The system is implemented as a web application, providing a user-friendly interface with features like pitch and rate adjustment for English. The system's performance is evaluated using Word Error Rate (WER) on diverse sentences, demonstrating its ability to handle various linguistic complexities within Igbo. This research contributes to enhancing accessibility for Igbo speakers, promoting language preservation, and advancing TTS technology for low-resource languages.
Keywords: Igbo TTS, Dialectal TTS, Low-Resource Language TTS, Multilingual TTS, TTS for Igbo.
|
DOI:
10.17148/IJARCCE.2025.14204
