Speech-to-Speech Translation Pipelines for Conversations in Low-Resource Languages
This work addresses translation quality for low-resource languages in community interpreting, but it is incremental as it focuses on optimizing existing pipeline components.
The study tackled speech-to-speech translation for low-resource languages like Turkish and Pashto to/from French by evaluating over 60 pipelines, identifying the best one for each direction and finding that component ranks are generally independent of the pipeline.
The popularity of automatic speech-to-speech translation for human conversations is growing, but the quality varies significantly depending on the language pair. In a context of community interpreting for low-resource languages, namely Turkish and Pashto to/from French, we collected fine-tuning and testing data, and compared systems using several automatic metrics (BLEU, COMET, and BLASER) and human assessments. The pipelines included automatic speech recognition, machine translation, and speech synthesis, with local models and cloud-based commercial ones. Some components have been fine-tuned on our data. We evaluated over 60 pipelines and determined the best one for each direction. We also found that the ranks of components are generally independent of the rest of the pipeline.