Fine-tuning on Clean Data for End-to-End Speech Translation: FBK @ IWSLT 2018
This work addresses speech translation for language processing applications, but it is incremental as it builds on existing methods with minor enhancements.
The paper tackled the problem of end-to-end English-German speech translation by fine-tuning a state-of-the-art LSTM and CNN model on cleaned data, resulting in a BLEU score improvement of 1.0 point on validation and best scores of 9.7 for a single model and 10.24 for an ensemble on test data.
This paper describes FBK's submission to the end-to-end English-German speech translation task at IWSLT 2018. Our system relies on a state-of-the-art model based on LSTMs and CNNs, where the CNNs are used to reduce the temporal dimension of the audio input, which is in general much higher than machine translation input. Our model was trained only on the audio-to-text parallel data released for the task, and fine-tuned on cleaned subsets of the original training corpus. The addition of weight normalization and label smoothing improved the baseline system by 1.0 BLEU point on our validation set. The final submission also featured checkpoint averaging within a training run and ensemble decoding of models trained during multiple runs. On test data, our best single model obtained a BLEU score of 9.7, while the ensemble obtained a BLEU score of 10.24.