Towards hate speech detection in low-resource languages: Comparing ASR to acoustic word embeddings on Wolof and Swahili
This addresses hate speech monitoring in low-resource languages, offering a more data-efficient alternative to ASR for real-world applications.
The paper tackled hate speech detection via keyword spotting in low-resource languages like Wolof and Swahili, comparing ASR to acoustic word embeddings (AWE). In controlled tests, ASR trained on five minutes of data outperformed AWE, but in real-world Swahili radio broadcasts, AWE with one minute of templates matched the performance of ASR trained on 30 hours of data.
We consider hate speech detection through keyword spotting on radio broadcasts. One approach is to build an automatic speech recognition (ASR) system for the target low-resource language. We compare this to using acoustic word embedding (AWE) models that map speech segments to a space where matching words have similar vectors. We specifically use a multilingual AWE model trained on labelled data from well-resourced languages to spot keywords in data in the unseen target language. In contrast to ASR, the AWE approach only requires a few keyword exemplars. In controlled experiments on Wolof and Swahili where training and test data are from the same domain, an ASR model trained on just five minutes of data outperforms the AWE approach. But in an in-the-wild test on Swahili radio broadcasts with actual hate speech keywords, the AWE model (using one minute of template data) is more robust, giving similar performance to an ASR system trained on 30 hours of labelled data.