Sentiment analysis (SA) is an important task because of its vital role in analyzing people’s opinions. However, existing research is solely based on the English language with limited work on low-resource languages. This study introduced a new multi-class Urdu dataset based on user reviews for sentiment analysis. This dataset is gathered from various domains such as food and beverages, movies and plays, software and apps, politics, and sports. Our proposed dataset contains 9312 reviews manually annotated by human experts into three classes: positive, negative and neutral. The main goal of this research study is to create a manually annotated dataset for Urdu sentiment analysis and to set baseline results using rule-based, machine learning (SVM, NB, Adabbost, MLP, LR and RF) and deep learning (CNN-1D, LSTM, Bi-LSTM, GRU and Bi-GRU) techniques. Additionally, we fine-tuned Multilingual BERT(mBERT) for Urdu sentiment analysis. We used four text representations: word n-grams, char n-grams,pre-trained fastText and BERT word embeddings to train our classifiers. We trained these models on two different datasets for evaluation purposes. Finding shows that the proposed mBERT model with BERT pre-trained word embeddings outperformed deep learning, machine learning and rule-based classifiers and achieved an F1 score of 81.49%.

Sentiment analysis (SA) has been an active research subject in the domain of natural language processing due to its important functions in interpreting people’s perspectives and drawing successful opinion-based judgments. On social media, Roman Urdu is one of the most extensively utilized dialects. Sentiment analysis of Roman Urdu is difficult due to its morphological complexities and varied dialects. The purpose of this paper is to evaluate the performance of various word embeddings for Roman Urdu and English dialects using the CNN-LSTM architecture with traditional machine learning classifiers. We introduce a novel deep learning architecture for Roman Urdu and English dialect SA based on two layers: LSTM for long-term dependency preservation and a one-layer CNN model for local feature extraction. To obtain the final classification, the feature maps learned by CNN and LSTM are fed to several machine learning classifiers. Various word embedding models support this concept. Extensive tests on four corpora show that the proposed model performs exceptionally well in Roman Urdu and English text sentiment classification, with an accuracy of 0.904, 0.841, 0.740, and 0.748 against MDPI, RUSA, RUSA-19, and UCL datasets, respectively. The results show that the SVM classifier and the Word2Vec CBOW (Continuous Bag of Words) model are more beneficial options for Roman Urdu sentiment analysis, but that BERT word embedding, two-layer LSTM, and SVM as a classifier function are more suitable options for English language sentiment analysis. The suggested model outperforms existing well-known advanced models on relevant corpora, improving the accuracy by up to 5%.

Although over 169 million people in the world are familiar with the Urdu language and a large quantity of Urdu data is being generated on different social websites daily, very few research studies and efforts have been completed to build language resources for the Urdu language and examine user sentiments. This study is focused on Urdu sentiment analysis of user reviews. After collecting Urdu user reviews about different genres from different websites, Urdu user reviews were annotated by three human experts. The primary objective of this study is twofold: (1) develop a benchmark dataset for resource-deprived Urdu language for sentiment analysis and (2) evaluate various machine learning and deep learning sentiment analysis models. Six machine learning and two deep learning classifiers, random forest (RF), naïve Bayes, support vector machine (SVM), AdaBoost, multilayer perceptron (MLP), logistic regression, LSTM, and CNN1D, are implemented. The results of all machine learning models are compared on the basis of different N-gram feature models. We implement all the above mentioned machine learning classifiers with unigram, bigram, trigram, uni-bigram, and uni-trigram features and deep learning models with the FastText word embedding model. Finally, the results of all classifiers are analyzed. The logistic regression model outperforms all other models in terms of accuracy of 0.8194, precision of 0.7995, recall of 0.8426, and an F1 measure of 0.8205 with the uni-trigram feature.

Speech emotion recognition (SER) is a challenging issue because it is not clear which features are effective for classification. Emotionally related features are always extracted from speech signals for emotional classification. Handcrafted features are mainly used for emotional identification from audio signals. However, these features are not sufficient to correctly identify the emotional state of the speaker. The advantages of a deep convolutional neural network (DCNN) are investigated in the proposed work. A pretrained framework is used to extract the features from speech emotion databases. In this work, we adopt the feature selection (FS) approach to find the discriminative and most important features for SER. Many algorithms are used for the emotion classification problem. We use the random forest (RF), decision tree (DT), support vector machine (SVM), multilayer perceptron classifier (MLP), and k-nearest neighbors (KNN) to classify seven emotions. All experiments are performed by utilizing four different publicly accessible databases. Our method obtains accuracies of 92.02%, 88.77%, 93.61%, and 77.23% for Emo-DB, SAVEE, RAVDESS, and IEMOCAP, respectively, for speaker-dependent (SD) recognition with the feature selection method. Furthermore, compared to current handcrafted feature-based SER methods, the proposed method shows the best results for speaker-independent SER. For EMO-DB, all classifiers attain an accuracy of more than 80% with or without the feature selection technique.

We needed to find deep emotional features to identify emotions from audio signals. Identifying emotions in spontaneous speech is a novel and challenging subject of research. Several CNN models were used to learn deep segment-level auditory representations of augmented Mel spectrograms. The proposed study introduces a novel technique for recognizing semi-natural and spontaneous speech emotions based on 1D (Model A) and 2D (Model B) deep convolutional neural networks (DCNNs) with two layers of long-short-term memory (LSTM). Both models used raw speech data and augmented (mid, left, right, and side) segment level Mel spectrograms to learn local and global features. The architecture of both models consists of five local feature learning blocks (LFLBs), two LSTM layers, and fully connected layers (FCL). In addition to learning local correlations and extracting hierarchical correlations, LFLB comprises two convolutional layers and a max-pooling layer. The LSTM layer learns long-term correlations from local features. The experiments illustrated that the proposed systems perform better than conventional methods. Model A achieved an average identification accuracy of 94.78% for speaker-dependent (SD) with a raw SAVEE dataset. With the IEMOCAP database, Model A achieved an average accuracy of an SD experiment with raw audio of 73.15%. In addition, Model A obtained identification accuracies of 97.19%, 94.09%, and 53.98% on SAVEE, IEMOCAP, and BAUM-1s, the databases for speaker-dependent (SD) experiments with an augmented Mel spectrogram, respectively. In contrast, Model B achieved identification accuracy of 96.85%, 88.80%, and 48.67% on SAVEE, IEMOCAP, and the BAUM-1s database for SI experiments with augmented reality Mel spectrogram, respectively.