AppTek's Submission to the IWSLT 2022 Isometric Spoken Language Translation Task

TL;DR

This paper describes AppTek's neural Transformer-based systems for English-German spoken language translation, focusing on length control mechanisms and data strategies to achieve high length compliance with minimal quality loss.

Contribution

The paper introduces novel length control techniques and data augmentation strategies that improve length compliance in spoken language translation systems.

Findings

Length compliance above 90% achieved

Minimal loss in BLEU and BERT scores

Effective use of synthetic and parallel data for quality

Abstract

To participate in the Isometric Spoken Language Translation Task of the IWSLT 2022 evaluation, constrained condition, AppTek developed neural Transformer-based systems for English-to-German with various mechanisms of length control, ranging from source-side and target-side pseudo-tokens to encoding of remaining length in characters that replaces positional encoding. We further increased translation length compliance by sentence-level selection of length-compliant hypotheses from different system variants, as well as rescoring of N-best candidates from a single system. Length-compliant back-translated and forward-translated synthetic data, as well as other parallel data variants derived from the original MuST-C training corpus were important for a good quality/desired length trade-off. Our experimental results show that length compliance levels above 90% can be reached while minimizing losses in MT quality as measured in BERT and BLEU scores.