Leveraging translations for speech transcription in low-resource settings
This work addresses transcription challenges for endangered language documentation, but it is incremental as it builds on existing data collection frameworks with a novel model variation.
The paper tackled the problem of improving speech transcription quality in low-resource endangered language settings by using text translations, and the result was a multi-source neural model that reduced transcription character error rate by up to 12.3% compared to baselines.
Recently proposed data collection frameworks for endangered language documentation aim not only to collect speech in the language of interest, but also to collect translations into a high-resource language that will render the collected resource interpretable. We focus on this scenario and explore whether we can improve transcription quality under these extremely low-resource settings with the assistance of text translations. We present a neural multi-source model and evaluate several variations of it on three low-resource datasets. We find that our multi-source model with shared attention outperforms the baselines, reducing transcription character error rate by up to 12.3%.