Ramon Sanabria

Ramon Sanabria, Hao Tang, Sharon Goldwater

Given the strong results of self-supervised models on various tasks, there have been surprisingly few studies exploring self-supervised representations for acoustic word embeddings (AWE), fixed-dimensional vectors representing variable-length spoken word segments. In this work, we study several pre-trained models and pooling methods for constructing AWEs with self-supervised representations. Owing to the contextualized nature of self-supervised representations, we hypothesize that simple pooling methods, such as averaging, might already be useful for constructing AWEs. When evaluating on a standard word discrimination task, we find that HuBERT representations with mean-pooling rival the state of the art on English AWEs. More surprisingly, despite being trained only on English, HuBERT representations evaluated on Xitsonga, Mandarin, and French consistently outperform the multilingual model XLSR-53 (as well as Wav2Vec 2.0 trained on English).

Ramon Sanabria, Wei-Ning Hsu, Alexei Baevski et al.

Human speech data comprises a rich set of domain factors such as accent, syntactic and semantic variety, or acoustic environment. Previous work explores the effect of domain mismatch in automatic speech recognition between pre-training and fine-tuning as a whole but does not dissect the contribution of individual factors. In this paper, we present a controlled study to better understand the effect of such factors on the performance of pre-trained representations on automatic speech recognition. To do so, we pre-train models either on modified natural speech or synthesized audio, with a single domain factor modified, and then measure performance after fine-tuning. Results show that phonetic domain factors play an important role during pre-training while grammatical and syntactic factors are far less important. To our knowledge, this is the first study to better understand the domain characteristics of pre-trained sets in self-supervised pre-training for speech.

Ramon Sanabria, Nikolay Bogoychev, Nina Markl et al.

English is the most widely spoken language in the world, used daily by millions of people as a first or second language in many different contexts. As a result, there are many varieties of English. Although the great many advances in English automatic speech recognition (ASR) over the past decades, results are usually reported based on test datasets which fail to represent the diversity of English as spoken today around the globe. We present the first release of The Edinburgh International Accents of English Corpus (EdAcc). This dataset attempts to better represent the wide diversity of English, encompassing almost 40 hours of dyadic video call conversations between friends. Unlike other datasets, EdAcc includes a wide range of first and second-language varieties of English and a linguistic background profile of each speaker. Results on latest public, and commercial models show that EdAcc highlights shortcomings of current English ASR models. The best performing model, trained on 680 thousand hours of transcribed data, obtains an average of 19.7% word error rate (WER) -- in contrast to the 2.7% WER obtained when evaluated on US English clean read speech. Across all models, we observe a drop in performance on Indian, Jamaican, and Nigerian English speakers. Recordings, linguistic backgrounds, data statement, and evaluation scripts are released on our website (https://groups.inf.ed.ac.uk/edacc/) under CC-BY-SA license.

Ramon Sanabria, Ondrej Klejch, Hao Tang et al.

Acoustic word embeddings are typically created by training a pooling function using pairs of word-like units. For unsupervised systems, these are mined using k-nearest neighbor (KNN) search, which is slow. Recently, mean-pooled representations from a pre-trained self-supervised English model were suggested as a promising alternative, but their performance on target languages was not fully competitive. Here, we explore improvements to both approaches: we use continued pre-training to adapt the self-supervised model to the target language, and we use a multilingual phone recognizer (MPR) to mine phone n-gram pairs for training the pooling function. Evaluating on four languages, we show that both methods outperform a recent approach on word discrimination. Moreover, the MPR method is orders of magnitude faster than KNN, and is highly data efficient. We also show a small improvement from performing learned pooling on top of the continued pre-trained representations.

Alexandra Saliba, Yuanchao Li, Ramon Sanabria et al.

The efficacy of self-supervised speech models has been validated, yet the optimal utilization of their representations remains challenging across diverse tasks. In this study, we delve into Acoustic Word Embeddings (AWEs), a fixed-length feature derived from continuous representations, to explore their advantages in specific tasks. AWEs have previously shown utility in capturing acoustic discriminability. In light of this, we propose measuring layer-wise similarity between AWEs and word embeddings, aiming to further investigate the inherent context within AWEs. Moreover, we evaluate the contribution of AWEs, in comparison to other types of speech features, in the context of Speech Emotion Recognition (SER). Through a comparative experiment and a layer-wise accuracy analysis on two distinct corpora, IEMOCAP and ESD, we explore differences between AWEs and raw self-supervised representations, as well as the proper utilization of AWEs alone and in combination with word embeddings. Our findings underscore the acoustic context conveyed by AWEs and showcase the highly competitive SER accuracies by appropriately employing AWEs.

Frank Palma Gomez, Ramon Sanabria, Yun-hsuan Sung et al. · cmu

Large language models (LLMs) are trained on text-only data that go far beyond the languages with paired speech and text data. At the same time, Dual Encoder (DE) based retrieval systems project queries and documents into the same embedding space and have demonstrated their success in retrieval and bi-text mining. To match speech and text in many languages, we propose using LLMs to initialize multi-modal DE retrieval systems. Unlike traditional methods, our system doesn't require speech data during LLM pre-training and can exploit LLM's multilingual text understanding capabilities to match speech and text in languages unseen during retrieval training. Our multi-modal LLM-based retrieval system is capable of matching speech and text in 102 languages despite only training on 21 languages. Our system outperforms previous systems trained explicitly on all 102 languages. We achieve a 10% absolute improvement in Recall@1 averaged across these languages. Additionally, our model demonstrates cross-lingual speech and text matching, which is further enhanced by readily available machine translation data.

Ramon Sanabria, Hao Tang, Sharon Goldwater

Word segmentation, the problem of finding word boundaries in speech, is of interest for a range of tasks. Previous papers have suggested that for sequence-to-sequence models trained on tasks such as speech translation or speech recognition, attention can be used to locate and segment the words. We show, however, that even on monolingual data this approach is brittle. In our experiments with different input types, data sizes, and segmentation algorithms, only models trained to predict phones from words succeed in the task. Models trained to predict words from either phones or speech (i.e., the opposite direction needed to generalize to new data), yield much worse results, suggesting that attention-based segmentation is only useful in limited scenarios.

Ramon Sanabria, Austin Waters, Jason Baldridge

Speech-based image retrieval has been studied as a proxy for joint representation learning, usually without emphasis on retrieval itself. As such, it is unclear how well speech-based retrieval can work in practice -- both in an absolute sense and versus alternative strategies that combine automatic speech recognition (ASR) with strong text encoders. In this work, we extensively study and expand choices of encoder architectures, training methodology (including unimodal and multimodal pretraining), and other factors. Our experiments cover different types of speech in three datasets: Flickr Audio, Places Audio, and Localized Narratives. Our best model configuration achieves large gains over state of the art, e.g., pushing recall-at-one from 21.8% to 33.2% for Flickr Audio and 27.6% to 53.4% for Places Audio. We also show our best speech-based models can match or exceed cascaded ASR-to-text encoding when speech is spontaneous, accented, or otherwise hard to automatically transcribe.

Tejas Srinivasan, Ramon Sanabria, Florian Metze et al.

Visual context has been shown to be useful for automatic speech recognition (ASR) systems when the speech signal is noisy or corrupted. Previous work, however, has only demonstrated the utility of visual context in an unrealistic setting, where a fixed set of words are systematically masked in the audio. In this paper, we simulate a more realistic masking scenario during model training, called RandWordMask, where the masking can occur for any word segment. Our experiments on the Flickr 8K Audio Captions Corpus show that multimodal ASR can generalize to recover different types of masked words in this unstructured masking setting. Moreover, our analysis shows that our models are capable of attending to the visual signal when the audio signal is corrupted. These results show that multimodal ASR systems can leverage the visual signal in more generalized noisy scenarios.

Tejas Srinivasan, Ramon Sanabria, Florian Metze et al.

Multimodal automatic speech recognition systems integrate information from images to improve speech recognition quality, by grounding the speech in the visual context. While visual signals have been shown to be useful for recovering entities that have been masked in the audio, these models should be capable of recovering a broader range of word types. Existing systems rely on global visual features that represent the entire image, but localizing the relevant regions of the image will make it possible to recover a larger set of words, such as adjectives and verbs. In this paper, we propose a model that uses finer-grained visual information from different parts of the image, using automatic object proposals. In experiments on the Flickr8K Audio Captions Corpus, we find that our model improves over approaches that use global visual features, that the proposals enable the model to recover entities and other related words, such as adjectives, and that improvements are due to the model's ability to localize the correct proposals.

Tejas Srinivasan, Ramon Sanabria, Florian Metze

Speech is understood better by using visual context; for this reason, there have been many attempts to use images to adapt automatic speech recognition (ASR) systems. Current work, however, has shown that visually adapted ASR models only use images as a regularization signal, while completely ignoring their semantic content. In this paper, we present a set of experiments where we show the utility of the visual modality under noisy conditions. Our results show that multimodal ASR models can recover words which are masked in the input acoustic signal, by grounding its transcriptions using the visual representations. We observe that integrating visual context can result in up to 35% relative improvement in masked word recovery. These results demonstrate that end-to-end multimodal ASR systems can become more robust to noise by leveraging the visual context.

Tejas Srinivasan, Ramon Sanabria, Florian Metze

In Neural Machine Translation (NMT) the usage of subwords and characters as source and target units offers a simple and flexible solution for translation of rare and unseen words. However, selecting the optimal subword segmentation involves a trade-off between expressiveness and flexibility, and is language and dataset-dependent. We present Block Multitask Learning (BMTL), a novel NMT architecture that predicts multiple targets of different granularities simultaneously, removing the need to search for the optimal segmentation strategy. Our multi-task model exhibits improvements of up to 1.7 BLEU points on each decoder over single-task baseline models with the same number of parameters on datasets from two language pairs of IWSLT15 and one from IWSLT19. The multiple hypotheses generated at different granularities can be combined as a post-processing step to give better translations, which improves over hypothesis combination from baseline models while using substantially fewer parameters.

Tejas Srinivasan, Ramon Sanabria, Florian Metze

Multimodal learning allows us to leverage information from multiple sources (visual, acoustic and text), similar to our experience of the real world. However, it is currently unclear to what extent auxiliary modalities improve performance over unimodal models, and under what circumstances the auxiliary modalities are useful. We examine the utility of the auxiliary visual context in Multimodal Automatic Speech Recognition in adversarial settings, where we deprive the models from partial audio signal during inference time. Our experiments show that while MMASR models show significant gains over traditional speech-to-text architectures (upto 4.2% WER improvements), they do not incorporate visual information when the audio signal has been corrupted. This shows that current methods of integrating the visual modality do not improve model robustness to noise, and we need better visually grounded adaptation techniques.

Yasufumi Moriya, Ramon Sanabria, Florian Metze et al.

A vast amount of audio-visual data is available on the Internet thanks to video streaming services, to which users upload their content. However, there are difficulties in exploiting available data for supervised statistical models due to the lack of labels. Unfortunately, generating labels for such amount of data through human annotation can be expensive, time-consuming and prone to annotation errors. In this paper, we propose a method to automatically extract entity-video frame pairs from a collection of instruction videos by using speech transcriptions and videos. We conduct experiments on image recognition and visual grounding tasks on the automatically constructed entity-video frame dataset of How2. The models will be evaluated on new manually annotated portion of How2 dev5 and val set and on the Flickr30k dataset. This work constitutes a first step towards meta-algorithms capable of automatically construct task-specific training sets.

Ozan Caglayan, Ramon Sanabria, Shruti Palaskar et al.

Humans are capable of processing speech by making use of multiple sensory modalities. For example, the environment where a conversation takes place generally provides semantic and/or acoustic context that helps us to resolve ambiguities or to recall named entities. Motivated by this, there have been many works studying the integration of visual information into the speech recognition pipeline. Specifically, in our previous work, we propose a multistep visual adaptive training approach which improves the accuracy of an audio-based Automatic Speech Recognition (ASR) system. This approach, however, is not end-to-end as it requires fine-tuning the whole model with an adaptation layer. In this paper, we propose novel end-to-end multimodal ASR systems and compare them to the adaptive approach by using a range of visual representations obtained from state-of-the-art convolutional neural networks. We show that adaptive training is effective for S2S models leading to an absolute improvement of 1.4% in word error rate. As for the end-to-end systems, although they perform better than baseline, the improvements are slightly less than adaptive training, 0.8 absolute WER reduction in single-best models. Using ensemble decoding, end-to-end models reach a WER of 15% which is the lowest score among all systems.

Ramon Sanabria, Ozan Caglayan, Shruti Palaskar et al.

In this paper, we introduce How2, a multimodal collection of instructional videos with English subtitles and crowdsourced Portuguese translations. We also present integrated sequence-to-sequence baselines for machine translation, automatic speech recognition, spoken language translation, and multimodal summarization. By making available data and code for several multimodal natural language tasks, we hope to stimulate more research on these and similar challenges, to obtain a deeper understanding of multimodality in language processing.

Ramon Sanabria, Florian Metze

In Automatic Speech Recognition it is still challenging to learn useful intermediate representations when using high-level (or abstract) target units such as words. For that reason, character or phoneme based systems tend to outperform word-based systems when just few hundreds of hours of training data are being used. In this paper, we first show how hierarchical multi-task training can encourage the formation of useful intermediate representations. We achieve this by performing Connectionist Temporal Classification at different levels of the network with targets of different granularity. Our model thus performs predictions in multiple scales for the same input. On the standard 300h Switchboard training setup, our hierarchical multi-task architecture exhibits improvements over single-task architectures with the same number of parameters. Our model obtains 14.0% Word Error Rate on the Eval2000 Switchboard subset without any decoder or language model, outperforming the current state-of-the-art on acoustic-to-word models.

Shruti Palaskar, Ramon Sanabria, Florian Metze

Transcription or sub-titling of open-domain videos is still a challenging domain for Automatic Speech Recognition (ASR) due to the data's challenging acoustics, variable signal processing and the essentially unrestricted domain of the data. In previous work, we have shown that the visual channel -- specifically object and scene features -- can help to adapt the acoustic model (AM) and language model (LM) of a recognizer, and we are now expanding this work to end-to-end approaches. In the case of a Connectionist Temporal Classification (CTC)-based approach, we retain the separation of AM and LM, while for a sequence-to-sequence (S2S) approach, both information sources are adapted together, in a single model. This paper also analyzes the behavior of CTC and S2S models on noisy video data (How-To corpus), and compares it to results on the clean Wall Street Journal (WSJ) corpus, providing insight into the robustness of both approaches.

Siddharth Dalmia, Ramon Sanabria, Florian Metze et al.

Techniques for multi-lingual and cross-lingual speech recognition can help in low resource scenarios, to bootstrap systems and enable analysis of new languages and domains. End-to-end approaches, in particular sequence-based techniques, are attractive because of their simplicity and elegance. While it is possible to integrate traditional multi-lingual bottleneck feature extractors as front-ends, we show that end-to-end multi-lingual training of sequence models is effective on context independent models trained using Connectionist Temporal Classification (CTC) loss. We show that our model improves performance on Babel languages by over 6% absolute in terms of word/phoneme error rate when compared to mono-lingual systems built in the same setting for these languages. We also show that the trained model can be adapted cross-lingually to an unseen language using just 25% of the target data. We show that training on multiple languages is important for very low resource cross-lingual target scenarios, but not for multi-lingual testing scenarios. Here, it appears beneficial to include large well prepared datasets.

Thomas Zenkel, Ramon Sanabria, Florian Metze et al.

This paper proposes a novel approach to create an unit set for CTC based speech recognition systems. By using Byte Pair Encoding we learn an unit set of an arbitrary size on a given training text. In contrast to using characters or words as units this allows us to find a good trade-off between the size of our unit set and the available training data. We evaluate both Crossword units, that may span multiple word, and Subword units. By combining this approach with decoding methods using a separate language model we are able to achieve state of the art results for grapheme based CTC systems.

Thomas Zenkel, Ramon Sanabria, Florian Metze et al.

Connectionist Temporal Classification has recently attracted a lot of interest as it offers an elegant approach to building acoustic models (AMs) for speech recognition. The CTC loss function maps an input sequence of observable feature vectors to an output sequence of symbols. Output symbols are conditionally independent of each other under CTC loss, so a language model (LM) can be incorporated conveniently during decoding, retaining the traditional separation of acoustic and linguistic components in ASR. For fixed vocabularies, Weighted Finite State Transducers provide a strong baseline for efficient integration of CTC AMs with n-gram LMs. Character-based neural LMs provide a straight forward solution for open vocabulary speech recognition and all-neural models, and can be decoded with beam search. Finally, sequence-to-sequence models can be used to translate a sequence of individual sounds into a word string. We compare the performance of these three approaches, and analyze their error patterns, which provides insightful guidance for future research and development in this important area.

Ramon Sanabria, Florian Metze, Fernando De La Torre

Speech is one of the most effective ways of communication among humans. Even though audio is the most common way of transmitting speech, very important information can be found in other modalities, such as vision. Vision is particularly useful when the acoustic signal is corrupted. Multi-modal speech recognition however has not yet found wide-spread use, mostly because the temporal alignment and fusion of the different information sources is challenging. This paper presents an end-to-end audiovisual speech recognizer (AVSR), based on recurrent neural networks (RNN) with a connectionist temporal classification (CTC) loss function. CTC creates sparse "peaky" output activations, and we analyze the differences in the alignments of output targets (phonemes or visemes) between audio-only, video-only, and audio-visual feature representations. We present the first such experiments on the large vocabulary IBM ViaVoice database, which outperform previously published approaches on phone accuracy in clean and noisy conditions.