Knowing the particular context associated with a conversation can help improving the performance of an automatic speech recognition (ASR) system. For example, if we are provided with a list of in-context words or phrases — such as the speaker’s contacts or recent song playlists — during inference, we can bias the recognition process towards this list. There are many works addressing contextual ASR; however, there is few publicly available real benchmark for evaluation, making it difficult to compare different solutions. To this end, we provide a corpus (“ConEC”) and baselines to evaluate contextual ASR approaches, grounded on real-world applications. The ConEC corpus is based on public-domain earnings calls (ECs) and associated supplementary materials, such as presentation slides, earnings news release as well as a list of meeting participants’ names and affiliations. We demonstrate that such real contexts are noisier than artificially synthesized contexts that contain the ground truth, yet they still make great room for future improvement of contextual ASR technology

A majority of language technologies are tailored for a small number of high-resource languages, while relatively many low-resource languages are neglected. One such group, Creole languages, have long been marginalized in academic study, though their speakers could benefit from machine translation (MT). These languages are predominantly used in much of Latin America, Africa and the Caribbean. We present the largest cumulative dataset to date for Creole language MT, including 14.5M unique Creole sentences with parallel translations—11.6M of which we release publicly, and the largest bitexts gathered to date for 41 languages—the first ever for 21. In addition, we provide MT models supporting all 41 Creole languages in 172 translation directions. Given our diverse dataset, we produce a model for Creole language MT exposed to more genre diversity then ever before, which outperforms a genre-specific Creole MT model on its own benchmark for 23 of 34 translation directions.

This paper presents JHU’s submissions to the IWSLT 2023 dialectal and low-resource track of Tunisian Arabic to English speech translation. The Tunisian dialect lacks formal orthography and abundant training data, making it challenging to develop effective speech translation (ST) systems. To address these challenges, we explore the integration of large pre-trained machine translation (MT) models, such as mBART and NLLB-200 in both end-to-end (E2E) and cascaded speech translation (ST) systems. We also improve the performance of automatic speech recognition (ASR) through the use of pseudo-labeling data augmentation and channel matching on telephone data. Finally, we combine our E2E and cascaded ST systems with Minimum Bayes-Risk decoding. Our combined system achieves a BLEU score of 21.6 and 19.1 on test2 and test3, respectively.

We describe the Johns Hopkins ACL 60-60 Speech Translation systems submitted to the IWSLT 2023 Multilingual track, where we were tasked to translate ACL presentations from English into 10 languages. We developed cascaded speech translation systems for both the constrained and unconstrained subtracks. Our systems make use of pre-trained models as well as domain-specific corpora for this highly technical evaluation-only task. We find that the specific technical domain which ACL presentations fall into presents a unique challenge for both ASR and MT, and we present an error analysis and an ACL-specific corpus we produced to enable further work in this area.

This paper details the Johns Hopkins speech translation (ST) system used in the IWLST2022 dialect speech translation task. Our system uses a cascade of automatic speech recognition (ASR) and machine translation (MT). We use a Conformer model for ASR systems and a Transformer model for machine translation. Surprisingly, we found that while using additional ASR training data resulted in only a negligible change in performance as measured by BLEU or word error rate (WER), aggressive text normalization improved BLEU more significantly. We also describe an approach, similar to back-translation, for improving performance using synthetic dialectal source text produced from source sentences in mismatched dialects.

Low-resource Multilingual Neural Machine Translation (MNMT) is typically tasked with improving the translation performance on one or more language pairs with the aid of high-resource language pairs. In this paper and we propose two simple search based curricula – orderings of the multilingual training data – which help improve translation performance in conjunction with existing techniques such as fine-tuning. Additionally and we attempt to learn a curriculum for MNMT from scratch jointly with the training of the translation system using contextual multi-arm bandits. We show on the FLORES low-resource translation dataset that these learned curricula can provide better starting points for fine tuning and improve overall performance of the translation system.

Large web-crawled corpora represent an excellent resource for improving the performance of Neural Machine Translation (NMT) systems across several language pairs. However, since these corpora are typically extremely noisy, their use is fairly limited. Current approaches to deal with this problem mainly focus on filtering using heuristics or single features such as language model scores or bi-lingual similarity. This work presents an alternative approach which learns weights for multiple sentence-level features. These feature weights which are optimized directly for the task of improving translation performance, are used to score and filter sentences in the noisy corpora more effectively. We provide results of applying this technique to building NMT systems using the Paracrawl corpus for Estonian-English and show that it beats strong single feature baselines and hand designed combinations. Additionally, we analyze the sensitivity of this method to different types of noise and explore if the learned weights generalize to other language pairs using the Maltese-English Paracrawl corpus.

The Mixer series of speech corpora were collected over several years, principally to support annual NIST evaluations of speaker recognition (SR) technologies. These evaluations focused on conversational speech over a variety of channels and recording conditions. One of the series, Mixer-6, added a new condition, read speech, to support basic scientific research on speaker characteristics, as well as technology evaluation. With read speech it is possible to make relatively precise measurements of phonetic events and features, which can be correlated with the performance of speaker recognition algorithms, or directly used in phonetic analysis of speaker variability. The read speech, as originally recorded, was adequate for large-scale evaluations (e.g., fixed-text speaker ID algorithms) but only marginally suitable for acoustic-phonetic studies. Numerous errors due largely to speaker behavior remained in the corpus, with no record of their locations or rate of occurrence. We undertook the effort to correct this situation with automatic methods supplemented by human listening and annotation. The present paper describes the tools and methods, resulting corrections, and some examples of the kinds of research studies enabled by these enhancements.

Translation of the output of automatic speech recognition (ASR) systems, also known as speech translation, has received a lot of research interest recently. This is especially true for programs such as DARPA BOLT which focus on improving spontaneous human-human conversation across languages. However, this research is hindered by the dearth of datasets developed for this explicit purpose. For Egyptian Arabic-English, in particular, no parallel speechtranscription-translation dataset exists in the same domain. In order to support research in speech translation, we introduce the Callhome Egyptian Arabic-English Speech Translation Corpus. This supplements the existing LDC corpus with four reference translations for each utterance in the transcripts. The result is a three-way parallel dataset of Egyptian Arabic Speech, transcriptions and English translations.

Research into the translation of the output of automatic speech recognition (ASR) systems is hindered by the dearth of datasets developed for that explicit purpose. For SpanishEnglish translation, in particular, most parallel data available exists only in vastly different domains and registers. In order to support research on cross-lingual speech applications, we introduce the Fisher and Callhome Spanish-English Speech Translation Corpus, supplementing existing LDC audio and transcripts with (a) ASR 1-best, lattice, and oracle output produced by the Kaldi recognition system and (b) English translations obtained on Amazon’s Mechanical Turk. The result is a four-way parallel dataset of Spanish audio, transcriptions, ASR lattices, and English translations of approximately 38 hours of speech, with defined training, development, and held-out test sets. We conduct baseline machine translation experiments using models trained on the provided training data, and validate the dataset by corroborating a number of known results in the field, including the utility of in-domain (information, conversational) training data, increased performance translating lattices (instead of recognizer 1-best output), and the relationship between word error rate and BLEU score.

Discriminative training for MT usually involves numerous features and requires large-scale training set to reach reliable parameter estimation. Other than using the expensive human-labeled parallel corpora for training, semi-supervised methods have been proposed to generate huge amount of “hallucinated” data which relieves the data sparsity problem. However the large training set contains both good samples which are suitable for training and bad ones harmful to the training. How to select training samples from vast amount of data can greatly affect the training performance. In this paper we propose a method for selecting samples that are most suitable for discriminative training according to a criterion measuring the dataset quality. Our experimental results show that by adding samples to the training set selectively, we are able to exceed the performance of system trained with the same amount of samples selected randomly.

We extend discriminative n-gram language modeling techniques originally proposed for automatic speech recognition to a statistical machine translation task. In this context, we propose a novel data selection method that leads to good models using a fraction of the training data. We carry out systematic experiments on several benchmark tests for Chinese to English translation using a hierarchical phrase-based machine translation system, and show that a discriminative language model significantly improves upon a state-of-the-art baseline. The experiments also highlight the benefits of our data selection method.