Or Haim Anidjar

Or Haim Anidjar, Revital Marbel, Roi Yozevitch

Approaching Speech-to-Text and Automatic Speech Recognition problems in low-resource languages is notoriously challenging due to the scarcity of validated datasets and the diversity of dialects. Arabic, Russian, and Portuguese exemplify these difficulties, being low-resource languages due to the many dialects of these languages across different continents worldwide. Moreover, the variety of accents and pronunciations of such languages complicate ASR models' success. With the increasing popularity of Deep Learning and Transformers, acoustic models like the renowned Wav2Vec2 have achieved superior performance in the Speech Recognition field compared to state-of-the-art approaches. However, despite Wav2Vec2's improved efficiency over traditional methods, its performance significantly declines for under-represented languages, even though it requires significantly less labeled data. This paper introduces an end-to-end framework that enhances ASR systems fine-tuned on Wav2Vec2 through data augmentation techniques. To validate our framework's effectiveness, we conducted a detailed experimental evaluation using three datasets from Mozilla's Common Voice project in Arabic, Russian, and Portuguese. Additionally, the framework presented in this paper demonstrates robustness to different diacritics. Ultimately, our approach outperforms two previous baseline models, which are the pre-trained Wav2Vec2 and the well-known Whisper ASR model, resulting in an average relative improvement of 33.9\% in Word Error Rate and a 53.2\% relative improvement in Character Error Rate.

Or Haim Anidjar, Roi Yozevitch

In this research, we advanced a spoken language recognition system, moving beyond traditional feature vector-based models. Our improvements focused on effectively capturing language characteristics over extended periods using a specialized pooling layer. We utilized a broad dataset range from Common-Voice, targeting ten languages across Indo-European, Semitic, and East Asian families. The major innovation involved optimizing the architecture of Time Delay Neural Networks. We introduced additional layers and restructured these networks into a funnel shape, enhancing their ability to process complex linguistic patterns. A rigorous grid search determined the optimal settings for these networks, significantly boosting their efficiency in language pattern recognition from audio samples. The model underwent extensive training, including a phase with augmented data, to refine its capabilities. The culmination of these efforts is a highly accurate system, achieving a 97\% accuracy rate in language recognition. This advancement represents a notable contribution to artificial intelligence, specifically in improving the accuracy and efficiency of language processing systems, a critical aspect in the engineering of advanced speech recognition technologies.

O. H. Anidjar, C. Hajaj, A. Dvir et al.

Speaker Diarization (SD) consists of splitting or segmenting an input audio burst according to speaker identities. In this paper, we focus on the crucial task of the SD problem which is the audio segmenting process and suggest a solution for the Change Point Detection (CPD) problem. We empirically demonstrate the negative correlation between an increase in the number of speakers and the Recall and F1-Score measurements. This negative correlation is shown to be the outcome of a massive experimental evaluation process, which accounts its superiority to recently developed voice based solutions. In order to overcome the number of speakers issue, we suggest a robust solution based on a novel Natural Language Processing (NLP) technique, as well as a metadata features extraction process, rather than a vocal based alone. To the best of our knowledge, we are the first to propose an intelligent NLP based solution that (I) tackles the CPD problem with a dataset in Hebrew, and (II) solves the CPD variant of the SD problem. We empirically show, based on two distinct datasets, that our method is abled to accurately identify the CPDs in an audio burst with 82.12% and 89.02% of success in the Recall and F1-score measurements.