Offiong Bassey

Hi 👋, I'm Offiong Bassey, a machine learning engineer and researcher specializing in Natural Language Processing (NLP). My work spans audio processing, speech synthesis, machine translation, and real-time conversational AI, with a particular focus on low-resource and multilingual languages.

My research explores how multilingual models can perform efficiently in low-resource settings, especially under limited data conditions. At Plotweaver, where I currently work as a Machine Learning Engineer, I apply this to improving the scalability and efficiency of language models for practical, real-world applications.

Outside of work, I enjoy reading, watching movies, singing classical choral music, playing the trumpet, and listening to podcasts on AI and technology.

offiong

News

June 2026

Paper accepted to Interspeech 2026 - Sydney, Australia.

Mar 2026

Paper (Adapting Foundational ASR Models to Efik) accepted to Speakable @ LREC 2026 - Spain.

Feb 2026

Awarded full sponsorship to attend the AfricaNLP Workshop @ EACL 2026, Rabat, Morocco.

Feb 2026

Joined PlotWeaver as a Machine Learning Research Engineer.

Jan 2026

Gave a talk @ Kabod Group on Funding opportunities for African Language Industry Projects.

Jan 2026

First paper accepted to AfricaNLP @ EACL 2026 - Morocco.


Selected Publications

All Publications

Interspeech 2026

Towards Digital Preservation of Efik: TTS for a Low-Resource African Language

We curated a high-quality Efik speech dataset, trained and evaluated the performance of four neural Text-to-Speech (TTS) models for Efik under low-resource settings, with the goal of advancing speech technology and contributing to the digital preservation of the Efik language.

Speakable @ LREC 2026

Adapting Foundational ASR Models to Efik: An Empirical Study of an Extremely Low-Resource Tonal Language.

This paper investigates the adoption of state-of-the-art foundational ASR models such as XLS-R and Whisper through fine-tuning for Efik, a low-resource tonal language and empirically evaluates their performance.

AfricaNLP @ ACL 2026

Developing an English–Efik Corpus and Machine Translation System for Digital Inclusion.

This study evaluates the effectiveness of state-of-the-art multilingual neural machine translation models for English–Efik translation, leveraging a small-scale, community-curated parallel corpus of 13,865 sentence pairs.