Using Self-Supervised Learning (SSL) as model initialization is now common to obtain strong results in Speech Translation (ST). However, they also impose a large memory footprint, hindering on-device deployment. In this paper, we leverage the SSL models by pretraining smaller models on their Discrete Speech Units (DSU). We pretrain encoder-decoder models on 1) Filterbank-to-DSU and 2) DSU-to-Translation data, and take the encoder from 1) and the decoder from 2) to initialise a new model, finetuning this on limited speech-translation data. The final model becomes compact by using the DSU pretraining to distil the knowledge of the SSL model. Our method has several benefits over using DSU as model inputs, such as shorter inference pipeline and robustness over (DSU) tokenization. In contrast to ASR pretraining, it does not require transcripts, making it applicable to low-resource settings. Evaluation on CoVoST-2 X-En shows that our method is >$0.5$ BLEU better than a ST model that directly finetune the SSL model, given only half the model size, and on a par with ASR pretraining.
Text continues to remain a relevant form of representation for information. Text documents are created either in digital native platforms or through the conversion of other media files such as images and speech. While the digital native text is invariably obtained through physical or virtual keyboards, technologies such as OCR and speech recognition are utilized to transform the images and speech signals into text content. All these variety of mechanisms of text generation also introduce errors into the captured text. This project aims at analyzing different kinds of error that occurs in text documents. The work employs two of the advanced deep neural network-based language models, namely, BART and MarianMT, to rectify the anomalies present in the text. Transfer learning of these models with available dataset is performed to finetune their capacity for error correction. A comparative study is conducted to investigate the effectiveness of these models in handling each of the defined error categories. It is observed that while both models can bring down the erroneous sentences by 20+%, BART can handle spelling errors far better (24.6%) than grammatical errors (8.8%).
Lexical-syntactic flexibility, in the form of conversion (or zero-derivation) is a hallmark of English morphology. In conversion, a word with one part of speech is placed in a non-prototypical context, where it is coerced to behave as if it had a different part of speech. However, while this process affects a large part of the English lexicon, little work has been done to establish the degree to which language models capture this type of generalization. This paper reports the first study on the behavior of large language models with reference to conversion. We design a task for testing lexical-syntactic flexibility -- the degree to which models can generalize over words in a construction with a non-prototypical part of speech. This task is situated within a natural language inference paradigm. We test the abilities of five language models -- two proprietary models (GPT-3.5 and GPT-4), three open-source models (Mistral 7B, Falcon 40B, and Llama 2 70B). We find that GPT-4 performs best on the task, followed by GPT-3.5, but that the open source language models are also able to perform it and that the 7B parameter Mistral displays as little difference between its baseline performance on the natural language inference task and the non-prototypical syntactic category task, as the massive GPT-4.
The widespread use of automated voice assistants along with other recent technological developments have increased the demand for applications that process audio signals and human voice in particular. Voice recognition tasks are typically performed using artificial intelligence and machine learning models. Even though end-to-end models exist, properly pre-processing the signal can greatly reduce the complexity of the task and allow it to be solved with a simpler ML model and fewer computational resources. However, ML engineers who work on such tasks might not have a background in signal processing which is an entirely different area of expertise. The objective of this work is to provide a concise comparative analysis of Fourier and Wavelet transforms that are most commonly used as signal decomposition methods for audio processing tasks. Metrics for evaluating speech intelligibility are also discussed, namely Scale-Invariant Signal-to-Distortion Ratio (SI-SDR), Perceptual Evaluation of Speech Quality (PESQ), and Short-Time Objective Intelligibility (STOI). The level of detail in the exposition is meant to be sufficient for an ML engineer to make informed decisions when choosing, fine-tuning, and evaluating a decomposition method for a specific ML model. The exposition contains mathematical definitions of the relevant concepts accompanied with intuitive non-mathematical explanations in order to make the text more accessible to engineers without deep expertise in signal processing. Formal mathematical definitions and proofs of theorems are intentionally omitted in order to keep the text concise.
In today's digital world, cyberbullying is a serious problem that can harm the mental and physical health of people who use social media. This paper explains just how serious cyberbullying is and how it really affects indi-viduals exposed to it. It also stresses how important it is to find better ways to detect cyberbullying so that online spaces can be safer. Plus, it talks about how making more accurate tools to spot cyberbullying will be really helpful in the future. Our paper introduces a deep learning-based ap-proach, primarily employing BERT and BiLSTM architectures, to effective-ly address cyberbullying. This approach is designed to analyse large vol-umes of posts and predict potential instances of cyberbullying in online spaces. Our results demonstrate the superiority of the hateBERT model, an extension of BERT focused on hate speech detection, among the five mod-els, achieving an accuracy rate of 89.16%. This research is a significant con-tribution to "Computational Intelligence for Social Transformation," prom-ising a safer and more inclusive digital landscape.
Large language models (LLMs) excel in many diverse applications beyond language generation, e.g., translation, summarization, and sentiment analysis. One intriguing application is in text classification. This becomes pertinent in the realm of identifying hateful or toxic speech -- a domain fraught with challenges and ethical dilemmas. In our study, we have two objectives: firstly, to offer a literature review revolving around LLMs as classifiers, emphasizing their role in detecting and classifying hateful or toxic content. Subsequently, we explore the efficacy of several LLMs in classifying hate speech: identifying which LLMs excel in this task as well as their underlying attributes and training. Providing insight into the factors that contribute to an LLM proficiency (or lack thereof) in discerning hateful content. By combining a comprehensive literature review with an empirical analysis, our paper strives to shed light on the capabilities and constraints of LLMs in the crucial domain of hate speech detection.
Different fields in applied machine learning such as computer vision, speech or natural language processing have been building domain-specialised solutions. Currently, we are witnessing an opposing trend towards developing more generalist architectures, driven by Large Language Models and multi-modal foundational models. These architectures are designed to tackle a variety of tasks, including those previously unseen and using inputs across multiple modalities. Taking this trend of generalization to the extreme suggests the possibility of a single deep network architecture capable of solving all tasks. This position paper aims to explore developing such a unified architecture and proposes a theoretical framework of how it could be constructed. Our proposal is based on the following assumptions. Firstly, tasks are solved by following a sequence of instructions, typically implemented in code for conventional computing hardware, which inherently operates sequentially. Second, recent Generative AI, especially Transformer-based models, demonstrate potential as an architecture capable of constructing algorithms for a wide range of domains. For example, GPT-4 shows exceptional capability at in-context learning of novel tasks which is hard to explain in any other way than the ability to compose novel solutions from fragments on previously learnt algorithms. Third, the observation that the main missing component in developing a truly generalised network is an efficient approach for self-consistent input of previously learnt sub-steps of an algorithm and their (implicit) composition during the network's internal forward pass. Our exploration delves into current capabilities and limitations of Transformer-based and other methods in efficient and correct algorithm composition and proposes a Transformer-like architecture as well as a discrete learning framework to overcome these limitations.
It remains a significant challenge how to quantitatively control the expressiveness of speech emotion in speech generation. In this work, we present a novel approach for manipulating the rendering of emotions for speech generation. We propose a hierarchical emotion distribution extractor, i.e. Hierarchical ED, that quantifies the intensity of emotions at different levels of granularity. Support vector machines (SVMs) are employed to rank emotion intensity, resulting in a hierarchical emotional embedding. Hierarchical ED is subsequently integrated into the FastSpeech2 framework, guiding the model to learn emotion intensity at phoneme, word, and utterance levels. During synthesis, users can manually edit the emotional intensity of the generated voices. Both objective and subjective evaluations demonstrate the effectiveness of the proposed network in terms of fine-grained quantitative emotion editing.
This work is an attempt to introduce a comprehensive benchmark for Arabic speech recognition, specifically tailored to address the challenges of telephone conversations in Arabic language. Arabic, characterized by its rich dialectal diversity and phonetic complexity, presents a number of unique challenges for automatic speech recognition (ASR) systems. These challenges are further amplified in the domain of telephone calls, where audio quality, background noise, and conversational speech styles negatively affect recognition accuracy. Our work aims to establish a robust benchmark that not only encompasses the broad spectrum of Arabic dialects but also emulates the real-world conditions of call-based communications. By incorporating diverse dialectical expressions and accounting for the variable quality of call recordings, this benchmark seeks to provide a rigorous testing ground for the development and evaluation of ASR systems capable of navigating the complexities of Arabic speech in telephonic contexts. This work also attempts to establish a baseline performance evaluation using state-of-the-art ASR technologies.
Kurdish, an Indo-European language spoken by over 30 million speakers, is considered a dialect continuum and known for its diversity in language varieties. Previous studies addressing language and speech technology for Kurdish handle it in a monolithic way as a macro-language, resulting in disparities for dialects and varieties for which there are few resources and tools available. In this paper, we take a step towards developing resources for language and speech technology for varieties of Central Kurdish, creating a corpus by transcribing movies and TV series as an alternative to fieldwork. Additionally, we report the performance of machine translation, automatic speech recognition, and language identification as downstream tasks evaluated on Central Kurdish varieties. Data and models are publicly available under an open license at https://github.com/sinaahmadi/CORDI.