Speech Emotion Recognition Tasks Research Articles

Hybrid data augmentation and deep attention-based dilated convolutional-recurrent neural networks for speech emotion recognition

Recently, speech emotion recognition (SER) has become an active research area in speech processing, particularly with the advent of deep learning (DL). Numerous DL-based methods have been proposed for SER. However, most of the existing DL-based models are complex and require a large amounts of data to achieve a good performance. In this study, a new framework of deep attention-based dilated convolutional-recurrent neural networks coupled with a hybrid data augmentation method was proposed for addressing SER tasks. The hybrid data augmentation method constitutes an upsampling technique for generating more speech data samples based on the traditional and generative adversarial network approaches. By leveraging both convolutional and recurrent neural networks in a dilated form along with an attention mechanism, the proposed DL framework can extract high-level representations from three-dimensional log Mel spectrogram features. Dilated convolutional neural networks acquire larger receptive fields, whereas dilated recurrent neural networks overcome complex dependencies as well as the vanishing and exploding gradient issues. Furthermore, the loss functions are reconfigured by combining the SoftMax loss and the center-based losses to classify various emotional states. The proposed framework was implemented using the Python programming language and the TensorFlow deep learning library. To validate the proposed framework, the EmoDB and ERC benchmark datasets, which are imbalanced and/or small datasets, were employed. The experimental results indicate that the proposed framework outperforms other related state-of-the-art methods, yielding the highest unweighted recall rates of 88.03 ± 1.39 (%) and 66.56 ± 0.67 (%) for the EmoDB and ERC datasets, respectively.

ACG-EmoCluster: A Novel Framework to Capture Spatial and Temporal Information from Emotional Speech Enhanced by DeepCluster

Speech emotion recognition (SER) is a task that tailors a matching function between the speech features and the emotion labels. Speech data have higher information saturation than images and stronger temporal coherence than text. This makes entirely and effectively learning speech features challenging when using feature extractors designed for images or texts. In this paper, we propose a novel semi-supervised framework for extracting spatial and temporal features from speech, called the ACG-EmoCluster. This framework is equipped with a feature extractor for simultaneously extracting the spatial and temporal features, as well as a clustering classifier for enhancing the speech representations through unsupervised learning. Specifically, the feature extractor combines an Attn–Convolution neural network and a Bidirectional Gated Recurrent Unit (BiGRU). The Attn–Convolution network enjoys a global spatial receptive field and can be generalized to the convolution block of any neural networks according to the data scale. The BiGRU is conducive to learning temporal information on a small-scale dataset, thereby alleviating data dependence. The experimental results on the MSP-Podcast demonstrate that our ACG-EmoCluster can capture effective speech representation and outperform all baselines in both supervised and semi-supervised SER tasks.

Speech Emotion Recognition Based on Swin-Transformer

The ability of machines to understand human subjective emotions is an essential link to realize artificial intelligence. How to extract and utilize information from audio signals is still a challenging task. By transforming acoustic signals into time-domain information represented by spectrograms, advanced algorithms in the field of computer vision can be applied to the field of acoustics. In this paper, we propose a Speech Emotion Recognition(SER) system based on Swin-Transformer(Swin). In addition to verifying the feasibility of Swin in SER task, we also compared the effectiveness of various spectrum maps under the same model parameters. Our model is validated on the IEMOCAP dataset and achieves competitive performance.

Survey of Deep Representation Learning for Speech Emotion Recognition

Traditionally, speech emotion recognition (SER) research has relied on manually handcrafted acoustic features using feature engineering. However, the design of handcrafted features for complex SER tasks requires significant manual eort, which impedes generalisability and slows the pace of innovation. This has motivated the adoption of representation learning techniques that can automatically learn an intermediate representation of the input signal without any manual feature engineering. Representation learning has led to improved SER performance and enabled rapid innovation. Its effectiveness has further increased with advances in deep learning (DL), which has facilitated \textit{deep representation learning} where hierarchical representations are automatically learned in a data-driven manner. This paper presents the first comprehensive survey on the important topic of deep representation learning for SER. We highlight various techniques, related challenges and identify important future areas of research. Our survey bridges the gap in the literature since existing surveys either focus on SER with hand-engineered features or representation learning in the general setting without focusing on SER.

An ensemble 1D-CNN-LSTM-GRU model with data augmentation for speech emotion recognition

Precise recognition of emotion from speech signals aids in enhancing human–computer interaction (HCI). The performance of a speech emotion recognition (SER) system depends on the derived features from speech signals. However, selecting the optimal set of feature representations remains the most challenging task in SER because the effectiveness of features varies with emotions. Most studies extract hidden local speech features ignoring the global long-term contextual representations of speech signals. The existing SER system suffers from low recognition performance mainly due to the scarcity of available data and sub-optimal feature representations. Motivated by the efficient feature extraction of convolutional neural network (CNN), long short-term memory (LSTM), and gated recurrent unit (GRU), this article proposes an ensemble utilizing the combined predictive performance of three different architectures. The first architecture uses 1D CNN followed by Fully Connected Networks (FCN). In the other two architectures, LSTM-FCN and GRU-FCN layers follow the CNN layer respectively. All three individual models focus on extracting both local and long-term global contextual representations of speech signals. The ensemble uses a weighted average of the individual models. We evaluated the model's performance on five benchmark datasets: TESS, EMO-DB, RAVDESS, SAVEE, and CREMA-D. We have augmented the data by injecting additive white gaussian noise, pitch shifting, and stretching the signal level to obtain better model generalization. Five categories of features were extracted from the speech samples: mel-frequency cepstral coefficients, log mel-scaled spectrogram, zero-crossing rate, chromagram, and root mean square value from each audio file in those datasets. All four models perform exceptionally well in the SER task; notably, the ensemble model accomplishes the state-of-the-art (SOTA) weighted average accuracy of 99.46% for TESS, 95.42% for EMO-DB, 95.62% for RAVDESS, 93.22% for SAVEE, and 90.47% for CREMA-D datasets and thus significantly outperformed the SOTA models using the same datasets.

Cross-Corpus Speech Emotion Recognition Based on Multi-Task Learning and Subdomain Adaptation.

To solve the problem of feature distribution discrepancy in cross-corpus speech emotion recognition tasks, this paper proposed an emotion recognition model based on multi-task learning and subdomain adaptation, which alleviates the impact on emotion recognition. Existing methods have shortcomings in speech feature representation and cross-corpus feature distribution alignment. The proposed model uses a deep denoising auto-encoder as a shared feature extraction network for multi-task learning, and the fully connected layer and softmax layer are added before each recognition task as task-specific layers. Subsequently, the subdomain adaptation algorithm of emotion and gender features is added to the shared network to obtain the shared emotion features and gender features of the source domain and target domain, respectively. Multi-task learning effectively enhances the representation ability of features, a subdomain adaptive algorithm promotes the migrating ability of features and effectively alleviates the impact of feature distribution differences in emotional features. The average results of six cross-corpus speech emotion recognition experiments show that, compared with other models, the weighted average recall rate is increased by 1.89~10.07%, the experimental results verify the validity of the proposed model.

Learning Speech Emotion Representations in the Quaternion Domain

The modeling of human emotion expression in speech signals is an important, yet challenging task. The high resource demand of speech emotion recognition models, combined with the general scarcity of emotion-labelled data are obstacles to the development and application of effective solutions in this field. In this paper, we present an approach to jointly circumvent these difficulties. Our method, named RH-emo, is a novel semi-supervised architecture aimed at extracting quaternion embeddings from real-valued monoaural spectrograms, enabling the use of quaternion-valued networks for speech emotion recognition tasks. RH-emo is a hybrid real/quaternion autoencoder network that consists of a real-valued encoder in parallel to a real-valued emotion classifier and a quaternion-valued decoder. On the one hand, the classifier permits to optimization of each latent axis of the embeddings for the classification of a specific emotion-related characteristic: valence, arousal, dominance, and overall emotion. On the other hand, quaternion reconstruction enables the latent dimension to develop intra-channel correlations that are required for an effective representation as a quaternion entity. We test our approach on speech emotion recognition tasks using four popular datasets: IEMOCAP, RAVDESS, EmoDB, and TESS, comparing the performance of three well-established real-valued CNN architectures (AlexNet, ResNet-50, VGG) and their quaternion-valued equivalent fed with the embeddings created with RH-emo. We obtain a consistent improvement in the test accuracy for all datasets, while drastically reducing the resources' demand of models. Moreover, we performed additional experiments and ablation studies that confirm the effectiveness of our approach. The RH-emo repository is available at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/ispamm/rhemo</uri> .

SpeechFormer++: A Hierarchical Efficient Framework for Paralinguistic Speech Processing

Paralinguistic speech processing is important in addressing many issues, such as sentiment and neurocognitive disorder analyses. Recently, Transformer has achieved remarkable success in the natural language processing field and has demonstrated its adaptation to speech. However, previous works on Transformer in the speech field have not incorporated the properties of speech, leaving the full potential of Transformer unexplored. In this paper, we consider the characteristics of speech and propose a general structure-based framework, called SpeechFormer++, for paralinguistic speech processing. More concretely, following the component relationship in the speech signal, we design a unit encoder to model the intra- and inter-unit information ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i> , frames, phones, and words) efficiently. According to the hierarchical relationship, we utilize merging blocks to generate features at different granularities, which is consistent with the structural pattern in the speech signal. Moreover, a word encoder is introduced to integrate word-grained features into each unit encoder, which effectively balances fine-grained and coarse-grained information. SpeechFormer++ is evaluated on the speech emotion recognition (IEMOCAP & MELD), depression classification (DAIC-WOZ) and Alzheimer's disease detection (Pitt) tasks. The results show that SpeechFormer++ outperforms the standard Transformer while greatly reducing the computational cost. Furthermore, it delivers superior results compared to the state-of-the-art approaches.

Speech Emotion Recognition Based on Parallel CNN-Attention Networks with Multi-Fold Data Augmentation

In this paper, an automatic speech emotion recognition (SER) task of classifying eight different emotions was experimented using parallel based networks trained using the Ryeson Audio-Visual Dataset of Speech and Song (RAVDESS) dataset. A combination of a CNN-based network and attention-based networks, running in parallel, was used to model both spatial features and temporal feature representations. Multiple Augmentation techniques using Additive White Gaussian Noise (AWGN), SpecAugment, Room Impulse Response (RIR), and Tanh Distortion techniques were used to augment the training data to further generalize the model representation. Raw audio data were transformed into Mel-Spectrograms as the model’s input. Using CNN’s proven capability in image classification and spatial feature representations, the spectrograms were treated as an image with the height and width represented by the spectrogram’s time and frequency scales. Temporal feature representations were represented by attention-based models Transformer, and BLSTM-Attention modules. Proposed architectures of the parallel CNN-based networks running along with Transformer and BLSTM-Attention modules were compared with standalone CNN architectures and attention-based networks, as well as with hybrid architectures with CNN layers wrapped in time-distributed wrappers stacked on attention-based networks. In these experiments, the highest accuracy of 89.33% for a Parallel CNN-Transformer network and 85.67% for a Parallel CNN-BLSTM-Attention Network were achieved on a 10% hold-out test set from the dataset. These networks showed promising results based on their accuracies, while keeping significantly less training parameters compared with non-parallel hybrid models.

An autoencoder-based feature level fusion for speech emotion recognition

Although speech emotion recognition is challenging, it has broad application prospects in human-computer interaction. Building a system that can accurately and stably recognize emotions from human languages can provide a better user experience. However, the current unimodal emotion feature representations are not distinctive enough to accomplish the recognition, and they do not effectively simulate the inter-modality dynamics in speech emotion recognition tasks. This paper proposes a multimodal method that utilizes both audio and semantic content for speech emotion recognition. The proposed method consists of three parts: two high-level feature extractors for text and audio modalities, and an autoencoder-based feature fusion. For audio modality, we propose a structure called Temporal Global Feature Extractor (TGFE) to extract the high-level features of the time-frequency domain relationship from the original speech signal. Considering that text lacks frequency information, we use only a Bidirectional Long Short-Term Memory network (BLSTM) and attention mechanism to simulate an intra-modal dynamic. Once these steps have been accomplished, the high-level text and audio features are sent to the autoencoder in parallel to learn their shared representation for final emotion classification. We conducted extensive experiments on three public benchmark datasets to evaluate our method. The results on Interactive Emotional Motion Capture (IEMOCAP) and Multimodal EmotionLines Dataset (MELD) outperform the existing method. Additionally, the results of CMU Multi-modal Opinion-level Sentiment Intensity (CMU-MOSI) are competitive. Furthermore, experimental results show that compared to unimodal information and autoencoder-based feature level fusion, the joint multimodal information (audio and text) improves the overall performance and can achieve greater accuracy than simple feature concatenation.

Rethinking Auditory Affective Descriptors Through Zero-Shot Emotion Recognition in Speech

Zero-shot speech emotion recognition (SER) endows machines with the ability of sensing unseen-emotional states in speech, compared with conventional SER endeavors on supervised cases. On addressing the zero-shot SER task, auditory affective descriptors (AADs) are typically employed to transfer affective knowledge from seen- to unseen-emotional states. However, it remains unknown which types of AADs can well describe emotional states in speech during the transfer. In this regard, we define and research on three types of AADs, namely, per-emotion semantic-embedding, per-emotion manually annotated, and per-sample manually annotated AADs, through zero-shot emotion recognition in speech. This leads to a systematic design including prototype- and annotation-based zero-shot SER modules, relying on the input from per-emotion and per-sample AADs, respectively. We then perform extensive experimental comparisons between human and machines’ AADs on the French emotional speech corpus CINEMO for positive-negative (PN) and within-negative (WN) tasks. The experimental results indicate that semantic-embedding prototypes from pretrained models can outperform manually annotated emotional dimensions in zero-shot SER. The results further demonstrate that it is possible for machines to understand and describe affective information in speech better than human beings, with the help of sufficient pretrained models.

GM-TCNet: Gated Multi-scale Temporal Convolutional Network using Emotion Causality for Speech Emotion Recognition

In human-computer interaction, Speech Emotion Recognition (SER) plays an essential role in understanding the user's intent and improving the interactive experience. While similar sentimental speeches own diverse speaker characteristics but share common antecedents and consequences, an essential challenge for SER is how to produce robust and discriminative representations through causality between speech emotions. In this paper, we propose a Gated Multi-scale Temporal Convolutional Network (GM-TCNet) to construct a novel emotional causality repre- sentation learning component with a multi-scale receptive field. GM-TCNet deploys a novel emotional causality representation learning component to capture the dynamics of emotion across the time domain, constructed with dilated causal convolutions layer and gating mechanism. Besides, it utilizes skip connection fusing high-level fea- tures from different Gated Convolution Blocks (GCB) to capture abundant and subtle emotion changes in human speech. GM-TCNet first uses a single type of feature, Mel-Frequency Cepstral Coefficients (MFCC), as inputs and then passes them through the Gated Temporal Convolutional Module (GTCM) to generate the high-level fea- tures. Finally, the features are fed to the emotion classifier to accomplish the SER task. The experimental results show that our model maintains the highest performance in most cases, with +0.90% to +18.50% and +0.55% to +20.15% average relative improvement on the weighted average recall and unweighted average recall compared to state-of-the-art techniques. The source code is available at: https://github.com/Jiaxin-Ye/GM-TCNet for SER.

Progressive distribution adapted neural networks for cross-corpus speech emotion recognition.

In this paper, we investigate a challenging but interesting task in the research of speech emotion recognition (SER), i.e., cross-corpus SER. Unlike the conventional SER, the training (source) and testing (target) samples in cross-corpus SER come from different speech corpora, which results in a feature distribution mismatch between them. Hence, the performance of most existing SER methods may sharply decrease. To cope with this problem, we propose a simple yet effective deep transfer learning method called progressive distribution adapted neural networks (PDAN). PDAN employs convolutional neural networks (CNN) as the backbone and the speech spectrum as the inputs to achieve an end-to-end learning framework. More importantly, its basic idea for solving cross-corpus SER is very straightforward, i.e., enhancing the backbone's corpus invariant feature learning ability by incorporating a progressive distribution adapted regularization term into the original loss function to guide the network training. To evaluate the proposed PDAN, extensive cross-corpus SER experiments on speech emotion corpora including EmoDB, eNTERFACE, and CASIA are conducted. Experimental results showed that the proposed PDAN outperforms most well-performing deep and subspace transfer learning methods in dealing with the cross-corpus SER tasks.

Adapting Multiple Distributions for Bridging Emotions from Different Speech Corpora.

In this paper, we focus on a challenging, but interesting, task in speech emotion recognition (SER), i.e., cross-corpus SER. Unlike conventional SER, a feature distribution mismatch may exist between the labeled source (training) and target (testing) speech samples in cross-corpus SER because they come from different speech emotion corpora, which degrades the performance of most well-performing SER methods. To address this issue, we propose a novel transfer subspace learning method called multiple distribution-adapted regression (MDAR) to bridge the gap between speech samples from different corpora. Specifically, MDAR aims to learn a projection matrix to build the relationship between the source speech features and emotion labels. A novel regularization term called multiple distribution adaption (MDA), consisting of a marginal and two conditional distribution-adapted operations, is designed to collaboratively enable such a discriminative projection matrix to be applicable to the target speech samples, regardless of speech corpus variance. Consequently, by resorting to the learned projection matrix, we are able to predict the emotion labels of target speech samples when only the source label information is given. To evaluate the proposed MDAR method, extensive cross-corpus SER tasks based on three different speech emotion corpora, i.e., EmoDB, eNTERFACE, and CASIA, were designed. Experimental results showed that the proposed MDAR outperformed most recent state-of-the-art transfer subspace learning methods and even performed better than several well-performing deep transfer learning methods in dealing with cross-corpus SER tasks.

Cross-Language Speech Emotion Recognition Using Bag-of-Word Representations, Domain Adaptation, and Data Augmentation.

To date, several methods have been explored for the challenging task of cross-language speech emotion recognition, including the bag-of-words (BoW) methodology for feature processing, domain adaptation for feature distribution “normalization”, and data augmentation to make machine learning algorithms more robust across testing conditions. Their combined use, however, has yet to be explored. In this paper, we aim to fill this gap and compare the benefits achieved by combining different domain adaptation strategies with the BoW method, as well as with data augmentation. Moreover, while domain adaptation strategies, such as the correlation alignment (CORAL) method, require knowledge of the test data language, we propose a variant that we term N-CORAL, in which test languages (in our case, Chinese) are mapped to a common distribution in an unsupervised manner. Experiments with German, French, and Hungarian language datasets were performed, and the proposed N-CORAL method, combined with BoW and data augmentation, was shown to achieve the best arousal and valence prediction accuracy, highlighting the usefulness of the proposed method for “in the wild” speech emotion recognition. In fact, N-CORAL combined with BoW was shown to provide robustness across languages, whereas data augmentation provided additional robustness against cross-corpus nuance factors.

A bimodal network based on Audio–Text-Interactional-Attention with ArcFace loss for speech emotion recognition

Speech emotion recognition (SER) is an essential part of human–computer interaction. Meanwhile, the SER has widely utilized multimodal information in SER in recent years. This paper focuses on exploiting the acoustic and textual modalities for the SER task. We propose a bimodal network based on an Audio–Text-Interactional-Attention (ATIA) structure which can facilitate the interaction and fusion of the emotionally salient information within the acoustic and textual modalities. We also explored four different ATIA structures and verified their effectiveness. Finally, we selected one ATIA structure to build our bimodal network with the best performance. Furthermore, our SER model adopts an additive angular margin loss, named ArcFace loss, applied to the deep face recognition field. Compared with the widespread Softmax loss, our visualization results demonstrated the effectiveness of the ArcFace loss function. ArcFace loss can improve the discriminate power of features by focusing on the angles between the features and the weights. As we know, it is the first time to apply ArcFace loss in the field of SER. Finally, the results show that the bimodal network combined ArcFace loss achieved 72.8% of Weighted Accuracy (WA) and 62.5% of Unweighted Accuracy (UA) for the seven-class emotion classification, and 82.4% of WA and 80.6% of UA for the four-class emotion classification on the IEMOCAP dataset.

LGCCT: A Light Gated and Crossed Complementation Transformer for Multimodal Speech Emotion Recognition.

Semantic-rich speech emotion recognition has a high degree of popularity in a range of areas. Speech emotion recognition aims to recognize human emotional states from utterances containing both acoustic and linguistic information. Since both textual and audio patterns play essential roles in speech emotion recognition (SER) tasks, various works have proposed novel modality fusing methods to exploit text and audio signals effectively. However, most of the high performance of existing models is dependent on a great number of learnable parameters, and they can only work well on data with fixed length. Therefore, minimizing computational overhead and improving generalization to unseen data with various lengths while maintaining a certain level of recognition accuracy is an urgent application problem. In this paper, we propose LGCCT, a light gated and crossed complementation transformer for multimodal speech emotion recognition. First, our model is capable of fusing modality information efficiently. Specifically, the acoustic features are extracted by CNN-BiLSTM while the textual features are extracted by BiLSTM. The modality-fused representation is then generated by the cross-attention module. We apply the gate-control mechanism to achieve the balanced integration of the original modality representation and the modality-fused representation. Second, the degree of attention focus can be considered, as the uncertainty and the entropy of the same token should converge to the same value independent of the length. To improve the generalization of the model to various testing-sequence lengths, we adopt the length-scaled dot product to calculate the attention score, which can be interpreted from a theoretical view of entropy. The operation of the length-scaled dot product is cheap but effective. Experiments are conducted on the benchmark dataset CMU-MOSEI. Compared to the baseline models, our model achieves an 81.0% F1 score with only 0.432 M parameters, showing an improvement in the balance between performance and the number of parameters. Moreover, the ablation study signifies the effectiveness of our model and its scalability to various input-sequence lengths, wherein the relative improvement is almost 20% of the baseline without a length-scaled dot product.

Cross-Corpus Speech Emotion Recognition Based on Joint Transfer Subspace Learning and Regression

Speech emotion recognition has become an attractive research topic due to various emotional states of speech signals in real-life scenarios. Most current speech emotion recognition methods are carried out on a single corpus. However, in practice, the training and testing data often come from different domains, e.g., different corpora. In this case, the model generalizability and recognition performance would decrease greatly due to the domain mismatch. To address this challenging problem, we present a transfer learning method, called joint transfer subspace learning and regression (JTSLR), for cross-corpus speech emotion recognition. Specifically, JTSLR performs transfer subspace learning and regression in a joint framework. First, we learn a latent subspace by introducing a discriminative maximum mean discrepancy (MMD) as the discrepancy metric. Then, we put forward a regression function in this latent subspace to describe the relationships between features and corresponding labels. Moreover, we present a label graph to help transfer knowledge from relevant source data to target data. Finally, we conduct extensive experiments on three popular emotional data sets. The results show that our method can outperform traditional methods and some state-of-the-art transfer learning algorithms for cross-corpus speech emotion recognition tasks.

Graph based emotion recognition with attention pooling for variable-length utterances

Previous speech emotion recognition (SER) methods normally deal with variable-length utterance inputs by padding shorter ones or clipping longer ones into equal-length utterances, which may introduce invalid information or discard useful emotional segments. To address this issue, in this paper, we cast the SER problem into a graph classification task by transforming variable-length utterances into graphs to avoid padding or cutting. In our approach, frames (short windowed segments) in an utterance are presented as nodes in a graph. Acoustic features extracted from frames are treated as node feature vectors and nodes are connected according to their temporal relationship. Different graph convolutional networks (GCNs) are explored for node/frame embedding learning, and kinds of graph pooling methods are compared to obtain graph/utterance-level emotional representation from node embeddings. Extensive experiments with different GCN components and pooling mechanisms are conducted on the IEMOCAP and MSP-IMPRO datasets. The experimental results show that a combination of GraphSAGE with multi-head attention pooling (MHAPool) achieves the best weighted accuracy (WA) and comparable unweighted accuracy (UA) on both datasets compared with other state-of-the-art SER models, which demonstrates the effectiveness of the proposed graph-based network for SER task.

Multi-Task Semi-Supervised Adversarial Autoencoding for Speech Emotion Recognition

Inspite the emerging importance of Speech Emotion Recognition (SER), the state-of-the-art accuracy is quite low and needs improvement to make commercial applications of SER viable. A key underlying reason for the low accuracy is the scarcity of emotion datasets, which is a challenge for developing any robust machine learning model in general. In this article, we propose a solution to this problem: a multi-task learning framework that uses auxiliary tasks for which data is abundantly available. We show that utilisation of this additional data can improve the primary task of SER for which only limited labelled data is available. In particular, we use gender identifications and speaker recognition as auxiliary tasks, which allow the use of very large datasets, e. g., speaker classification datasets. To maximise the benefit of multi-task learning, we further use an adversarial autoencoder (AAE) within our framework, which has a strong capability to learn powerful and discriminative features. Furthermore, the unsupervised AAE in combination with the supervised classification networks enables semi-supervised learning which incorporates a discriminative component in the AAE unsupervised training pipeline. This semi-supervised learning essentially helps to improve generalisation of our framework and thus leads to improvements in SER performance. The proposed model is rigorously evaluated for categorical and dimensional emotion, and cross-corpus scenarios. Experimental results demonstrate that the proposed model achieves state-of-the-art performance on two publicly available datasets.