Aligning Pre-trained Models for Spoken Language Translation

This paper investigates a novel approach to end-to-end speech translation (ST) based on aligning frozen pre-trained automatic speech recognition (ASR) and machine translation (MT) models via a small connector module (Q-Former, our Subsampler-Transformer Encoder). This connector bridges the gap between the speech and text modalities, transforming ASR encoder embeddings into the latent representation space of the MT encoder while being the only part of the system optimized during training. Experiments are conducted on the How2 English-Portuguese dataset as we investigate the alignment approach in a small-scale scenario focusing on ST. While keeping the size of the connector module constant and small in comparison ( < 5% of the size of the larger aligned models), increasing the size and capability of the foundation ASR and MT models universally improves translation results. We also find that the connectors can serve as domain adapters for the foundation MT models, significantly improving translation performance in the aligned ST setting. We conclude that this approach represents a viable and scalable approach to training end-to-end ST systems.

Improving Automatic Speech Recognition with Decoder-Centric Regularisation in Encoder-Decoder Models

This paper proposes a simple yet effective way of regularising the encoder-decoder-based automatic speech recognition (ASR) models that enhance the robustness of the model and improve the generalisation to out-of-domain scenarios. The proposed approach is dubbed as $\textbf{De}$coder-$\textbf{C}$entric $\textbf{R}$egularisation in $\textbf{E}$ncoder-$\textbf{D}$ecoder (DeCRED) architecture for ASR, where auxiliary classifier(s) is introduced in layers of the decoder module. Leveraging these classifiers, we propose two decoding strategies that re-estimate the next token probabilities. Using the recent E-branchformer architecture, we build strong ASR systems that obtained competitive WERs as compared to Whisper-medium and outperformed OWSM v3; while relying only on a fraction of training data and model size. On top of such a strong baseline, we show that DeCRED can further improve the results and, moreover, generalise much better to out-of-domain scenarios, where we show an absolute reduction of 2.7 and 2.9 WERs on AMI and Gigaspeech datasets, respectively. We provide extensive analysis and accompanying experiments that support the benefits of the proposed regularisation scheme.

Beyond the Labels: Unveiling Text-Dependency in Paralinguistic Speech Recognition Datasets

Paralinguistic traits like cognitive load and emotion are increasingly recognized as pivotal areas in speech recognition research, often examined through specialized datasets like CLSE and IEMOCAP. However, the integrity of these datasets is seldom scrutinized for text-dependency. This paper critically evaluates the prevalent assumption that machine learning models trained on such datasets genuinely learn to identify paralinguistic traits, rather than merely capturing lexical features. By examining the lexical overlap in these datasets and testing the performance of machine learning models, we expose significant text-dependency in trait-labeling. Our results suggest that some machine learning models, especially large pre-trained models like HuBERT, might inadvertently focus on lexical characteristics rather than the intended paralinguistic features. The study serves as a call to action for the research community to reevaluate the reliability of existing datasets and methodologies, ensuring that machine learning models genuinely learn what they are designed to recognize.

Strategies for improving low resource speech to text translation relying on pre-trained ASR models

This paper presents techniques and findings for improving the performance of low-resource speech to text translation (ST). We conducted experiments on both simulated and real-low resource setups, on language pairs English - Portuguese, and Tamasheq - French respectively. Using the encoder-decoder framework for ST, our results show that a multilingual automatic speech recognition system acts as a good initialization under low-resource scenarios. Furthermore, using the CTC as an additional objective for translation during training and decoding helps to reorder the internal representations and improves the final translation. Through our experiments, we try to identify various factors (initializations, objectives, and hyper-parameters) that contribute the most for improvements in low-resource setups. With only 300 hours of pre-training data, our model achieved 7.3 BLEU score on Tamasheq - French data, outperforming prior published works from IWSLT 2022 by 1.6 points.

Detecting English Speech in the Air Traffic Control Voice Communication

We launched a community platform for collecting the ATC speech world-wide in the ATCO2 project. Filtering out unseen non-English speech is one of the main components in the data processing pipeline. The proposed English Language Detection (ELD) system is based on the embeddings from Bayesian subspace multinomial model. It is trained on the word confusion network from an ASR system. It is robust, easy to train, and light weighted. We achieved 0.0439 equal-error-rate (EER), a 50% relative reduction as compared to the state-of-the-art acoustic ELD system based on x-vectors, in the in-domain scenario. Further, we achieved an EER of 0.1352, a 33% relative reduction as compared to the acoustic ELD, in the unseen language (out-of-domain) condition. We plan to publish the evaluation dataset from the ATCO2 project.

Rethinking the objectives of extractive question answering

This paper describes two generally applicable approaches towards the significant improvement of the performance of state-of-the-art extractive question answering (EQA) systems. Firstly, contrary to a common belief, it demonstrates that using the objective with independence assumption for span probability $P(a_s,a_e) = P(a_s)P(a_e)$ of span starting at position $a_s$ and ending at position $a_e$ may have adverse effects. Therefore we propose a new compound objective that models joint probability $P(a_s,a_e)$ directly, while still keeping the objective with independency assumption as an auxiliary objective. Our second approach shows the beneficial effect of distantly semi-supervised shared-normalization objective known from (Clark and Gardner, 2017). We show that normalizing over a set of documents similar to the golden passage, and marginalizing over all ground-truth answer string positions leads to the improvement of results from smaller statistical models. Our results are supported via experiments with three QA models (BidAF, BERT, ALBERT) over six datasets. The proposed approaches do not use any additional data. Our code, analysis, pretrained models, and individual results will be available online.

Bayesian multilingual topic model for zero-shot cross-lingual topic identification

This paper presents a Bayesian multilingual topic model for learning language-independent document embeddings. Our model learns to represent the documents in the form of Gaussian distributions, thereby encoding the uncertainty in its covariance. We propagate the learned uncertainties through linear classifiers for zero-shot cross-lingual topic identification. Our experiments on 5 language Europarl and Reuters (MLDoc) corpora show that the proposed model outperforms multi-lingual word embedding and BiLSTM sentence encoder based systems with significant margins in the majority of the transfer directions. Moreover, our system trained under a single day on a single GPU with much lower amounts of data performs competitively as compared to the state-of-the-art universal BiLSTM sentence encoder trained on 93 languages. Our experimental analysis shows that the amount of parallel data improves the overall performance of embeddings. Nonetheless, exploiting the uncertainties is always beneficial.

Learning document embeddings along with their uncertainties

Majority of the text modelling techniques yield only point estimates of document embeddings and lack in capturing the uncertainty of the estimates. These uncertainties give a notion of how well the embeddings represent a document. We present Bayesian subspace multinomial model (Bayesian SMM), a generative log-linear model that learns to represent documents in the form of Gaussian distributions, thereby encoding the uncertainty in its covariance. Additionally, in the proposed Bayesian SMM, we address a commonly encountered problem of intractability that appears during variational inference in mixed-logit models. We also present a generative Gaussian linear classifier for topic identification that exploits the uncertainty in document embeddings. Our intrinsic evaluation using perplexity measure shows that the proposed Bayesian SMM fits the data better as compared to variational auto-encoder based document model. Our topic identification experiments on speech (Fisher) and text (20Newsgroups) corpora show that the proposed Bayesian SMM is robust to over-fitting on unseen test data. The topic ID results show that the proposed model is significantly better than variational auto-encoder based methods and achieve similar results when compared to fully supervised discriminative models.

An Empirical Evaluation of Zero Resource Acoustic Unit Discovery

Acoustic unit discovery (AUD) is a process of automatically identifying a categorical acoustic unit inventory from speech and producing corresponding acoustic unit tokenizations. AUD provides an important avenue for unsupervised acoustic model training in a zero resource setting where expert-provided linguistic knowledge and transcribed speech are unavailable. Therefore, to further facilitate zero-resource AUD process, in this paper, we demonstrate acoustic feature representations can be significantly improved by (i) performing linear discriminant analysis (LDA) in an unsupervised self-trained fashion, and (ii) leveraging resources of other languages through building a multilingual bottleneck (BN) feature extractor to give effective cross-lingual generalization. Moreover, we perform comprehensive evaluations of AUD efficacy on multiple downstream speech applications, and their correlated performance suggests that AUD evaluations are feasible using different alternative language resources when only a subset of these evaluation resources can be available in typical zero resource applications.