Neural machine translation (NMT) has emerged recently as a promising statistical machine translation approach. In NMT, neural networks (NN) are directly used to produce translations, without relying on a pre-existing translation framework. In this work, we take a step towards bridging the gap between conventional word alignment models and NMT. We follow the hidden Markov model (HMM) approach that separates the alignment and lexical models. We propose a neural alignment model and combine it with a lexical neural model in a loglinear framework. The models are used in a standalone word-based decoder that explicitly hypothesizes alignments during search. We demonstrate that our system outperforms attention-based NMT on two tasks: IWSLT 2013 German→English and BOLT Chinese→English. We also show promising results for re-aligning the training data using neural models.

AppTek and RWTH Aachen University team together to participate in the offline and simultaneous speech translation tracks of IWSLT 2020. For the offline task, we create both cascaded and end-to-end speech translation systems, paying attention to careful data selection and weighting. In the cascaded approach, we combine high-quality hybrid automatic speech recognition (ASR) with the Transformer-based neural machine translation (NMT). Our endto-end direct speech translation systems benefit from pretraining of adapted encoder and decoder components, as well as synthetic data and fine-tuning and thus are able to compete with cascaded systems in terms of MT quality. For simultaneous translation, we utilize a novel architecture that makes dynamic decisions, learned from parallel data, to determine when to continue feeding on input or generate output words. Experiments with speech and text input show that even at low latency this architecture leads to superior translation results.

We propose a conversion of bilingual sentence pairs and the corresponding word alignments into novel linear sequences.These are joint translation and reordering (JTR) uniquely defined sequences, combining interdepending lexical and alignment dependencies on the word level into a single framework. They are constructed in a simple manner while capturing multiple alignments and empty words. JTR sequences can be used to train a variety of models. We investigate the performances of ngram models with modified Kneser-Ney smoothing, feed-forward and recurrent neural network architectures when estimated on JTR sequences, and compare them to the operation sequence model (Durrani et al., 2013b). Evaluations on the IWSLT German→English, WMT German→English and BOLT Chinese→English tasks show that JTR models improve state-of-the-art phrasebased systems by up to 2.2 BLEU.

This paper investigates the application of vector space models (VSMs) to the standard phrase-based machine translation pipeline. VSMs are models based on continuous word representations embedded in a vector space. We exploit word vectors to augment the phrase table with new inferred phrase pairs. This helps reduce out-of-vocabulary (OOV) words. In addition, we present a simple way to learn bilingually-constrained phrase vectors. The phrase vectors are then used to provide additional scoring of phrase pairs, which fits into the standard log-linear framework of phrase-based statistical machine translation. Both methods result in significant improvements over a competitive in-domain baseline applied to the Arabic-to-English task of IWSLT 2013.

Conditional Random Fields (CRFs) are a state-of-the-art approach to natural language processing tasks like grapheme-tophoneme (g2p) conversion which is used to produce pronunciations or pronunciation variants for almost all ASR pronunciation lexica. One drawback of CRFs is that for training, an alignment is needed between graphemes and phonemes, usually even 1-to-1. The quality of the g2p result heavily depends on this alignment. Since these alignments are usually not annotated within the corpora, external models have to be used to produce such an alignment in a preprocessing step. In this work, we propose two approaches to integrate the alignment generation directly and efficiently into the CRF training process. Whereas the first approach relies on linear segmentation as starting point, the second approach considers all possible alignments given certain constraints. Both methods have been evaluated on two English g2p tasks, namely NETtalk and Celex, on which state-of-the-art results have been reported in the literature. The proposed approaches lead to results comparable to the state-of-the art.