DONDO: Open w2v-BERT Speech-Recognition Base Models for African Languages
2026-07-23 • Computation and Language
Computation and Language
AI summaryⓘ
The authors introduce DONDO, a set of open-source speech recognition models designed for 27 African language varieties. They trained these models mostly on read speech from religious texts, which are well-documented and consistent. Their approach fine-tunes a shared multilingual model with a special learning method to improve accuracy and allows the model to switch languages using a simple language identifier. The models perform nearly as well as language-specific models but cover multiple languages at once. All models are freely available for use and further training.
automatic speech recognitionself-supervised learningw2v-BERTfine-tuningmultilingual modelsword error ratelanguage conditioningAfrican languagesopen-source modelsHugging Face
Authors
Paul Azunre
Abstract
We present DONDO, a family of open, permissively licensed automatic speech recognition (ASR) base models for African languages, built on the w2v-BERT 2.0 self-supervised speech encoder. DONDO comprises twenty-one monolingual models and five multilingual models spanning twenty-seven language varieties across Ghana, Sierra Leone, Nigeria, Senegal, Kenya and Zimbabwe. Models are fine-tuned primarily on read speech drawn from religious texts, which offer broad, license-clear and orthographically consistent coverage for languages that otherwise lack transcribed audio. We describe a two-step (and, for one family, three-step) learning-rate-annealed fine-tuning procedure that first adapts a shared multilingual model at a high learning rate and then anneals it to recover, and in several cases surpass, strong monolingual baselines. We further describe a lightweight language-conditioning mechanism that injects a one-hot language identity as a sequence of prefix frames prepended to the acoustic features, allowing a single multilingual checkpoint to be steered to a target language at inference. Across the five multilingual families the annealed models reach average word error rates (WER) of 10-13%, closing most of the gap to monolingual models while covering many languages in a single checkpoint. All models are released on the Hugging Face KhayaAI organisation under the Apache-2.0 license (attribution only) so that others may fine-tune them freely, including for commercial use. We provide a conservative estimate that the languages covered are spoken by on the order of one hundred million first-language speakers, and by substantially more when second-language use is included.