Automatic speech recognition improves Ghanaian health communication languages
Benchmarking and Domain Adaptation of Automatic Speech Recognition (ASR) for Adolescent Health Communication in Ghanaian Languages
Computation and LanguageArtificial Intelligence
Summary
The paper studies how well speech recognition technology works for adolescent health talks in three Ghanaian languages. The authors tested several speech recognition systems on general and health-specific recordings, finding that fine-tuning on in-domain data greatly improved accuracy. They also deployed a live app to confirm the improvements and found that gathering real-world, task-specific data was more important than more powerful models. This means making speech tech useful in these languages depends most on having enough relevant examples to learn from.
What this means in practice
- •For mobile app developers: Build voice-based adolescent health communication tools in Ghanaian languages using adapted speech recognition models.
- •For health communication teams: Deploy accurate speech recognition to enable voice-first health information services tailored for youth in Ghanaian linguistic communities.
Authors
Stephen E. Moore, Akwasi Asare, Mich-Seth Owusu, Paul Azunre, Joel Budu, Lawrence A. Adu-Gyamfi
Abstract
This paper presents an end-to-end study of automatic speech recognition (ASR) for adolescent health communication in three Ghanaian languages (Twi, Dagbani, and Ewe). The work proceeds in three connected stages; First, we benchmark five ASR systems (three language-specific Wav2Vec2 models and two multimodal LLMs, Gemma 3n and Gemma 4) on a general-domain Bible corpus and a Youth Adolescent Sexual and Reproductive Health (ASRH) Domain ASR dataset, using Character and Word Error Rate (CER, WER). Second, guided by the benchmark, we perform supervised domain adaptation: although Gemma 4 was the strongest zero-shot candidate, fine-tuning it proved computationally infeasible, so we pivoted to the compact Qwen3-ASR-0.6B, fine-tuned on a large Ghana Bible corpus (~90k samples) and evaluated strictly on held-out human-collected in-domain audio. Fine-tuning reduced WER on every language, most dramatically for Ewe (WER from 109.3% to 64.8%, a drop of 44.5 pp; CER from 65.1% to 24.9%). Third, we validate the work through KasaHealth, a live voice-first ASRH application deployed in all three languages, complemented by Senti-Check, a technical evaluation harness. KasaHealth was tested by 50 community respondents and achieved a 100% chat-approval rate, a 72% Good-or-Excellent translation rating, and a 92% would-recommend rate, while surfacing the domain gaps that most constrain real-world use. Across all three stages the evidence converges: for these languages the binding constraint is validated in-domain data, not model capability or computation.