Electrical Engineering And Systems Science

Recent advances in neural TTS have led to models that can produce high-quality synthetic speech. However, these models typically require large amounts of training data, which can make it costly to produce a new voice with the desired quality. Although multi-speaker modeling can reduce the data requirements necessary for a new voice, this approach is usually not viable for many low-resource languages for which abundant multi-speaker data is not available. In this paper, we therefore investigated to what extent multilingual multi-speaker modeling can be an alternative to monolingual multi-speaker modeling, and explored how data from foreign languages may best be combined with low-resource language data. We found that multilingual modeling can increase the naturalness of low-resource language speech, showed that multilingual models can produce speech with a naturalness comparable to monolingual multi-speaker models, and saw that the target language naturalness was affected by the strategy used to add foreign language data.

Human emotional speech is, by its very nature, a variant signal. This results in dynamics intrinsic to automatic emotion classification based on speech. In this work, we explore a spectral decomposition method stemming from fluid-dynamics, known as Dynamic Mode Decomposition (DMD), to computationally represent and analyze the global utterance-level dynamics of emotional speech. Specifically, segment-level emotion-specific representations are first learned through an Emotion Distillation process. This forms a multi-dimensional signal of emotion flow for each utterance, called Emotion Profiles (EPs). The DMD algorithm is then applied to the resultant EPs to capture the eigenfrequencies, and hence the fundamental transition dynamics of the emotion flow. Evaluation experiments using the proposed approach, which we call EigenEmo, show promising results. Moreover, due to the positive combination of their complementary properties, concatenating the utterance representations generated by EigenEmo with simple EPs averaging yields noticeable gains.

Traditional approaches to automatic emotion recognition are relying on the application of handcrafted features. More recently however the advent of deep learning enabled algorithms to learn meaningful representations of input data automatically. In this paper, we investigate the applicability of transferring knowledge learned from large text and audio corpora to the task of automatic emotion recognition. To evaluate the practicability of our approach, we are taking part in this year's Interspeech ComParE Elderly Emotion Sub-Challenge, where the goal is to classify spoken narratives of elderly people with respect to the emotion of the speaker. Our results show that the learned feature representations can be effectively applied for classifying emotions from spoken language. We found the performance of the features extracted from the audio signal to be not as consistent as those that have been extracted from the transcripts. While the acoustic features achieved best in class results on the development set, when compared to the baseline systems, their performance dropped considerably on the test set of the challenge. The features extracted from the text form, however, are showing promising results on both sets and are outperforming the official baseline by 5.7 percentage points unweighted average recall.

Emotional voice conversion models adapt the emotion in speech without changing the speaker identity or linguistic content. They are less data hungry than text-to-speech models and allow to generate large amounts of emotional data for downstream tasks. In this work we propose EmoCat, a language-agnostic emotional voice conversion model. It achieves high-quality emotion conversion in German with less than 45 minutes of German emotional recordings by exploiting large amounts of emotional data in US English. EmoCat is an encoder-decoder model based on CopyCat, a voice conversion system which transfers prosody. We use adversarial training to remove emotion leakage from the encoder to the decoder. The adversarial training is improved by a novel contribution to gradient reversal to truly reverse gradients. This allows to remove only the leaking information and to converge to better optima with higher conversion performance. Evaluations show that Emocat can convert to different emotions but misses on emotion intensity compared to the recordings, especially for very expressive emotions. EmoCat is able to achieve audio quality on par with the recordings for five out of six tested emotion intensities.

Human emotions are inherently ambiguous and impure. When designing systems to anticipate human emotions based on speech, the lack of emotional purity must be considered. However, most of the current methods for speech emotion classification rest on the consensus, e.g., one single hard label for an utterance. This labeling principle imposes challenges for system performance considering emotional impurity. In this paper, we recommend the use of emotional profiles (EPs), which provides a time series of segment-level soft labels to capture the subtle blends of emotional cues present across a specific speech utterance. We further propose the emotion profile refinery (EPR), an iterative procedure to update EPs. The EPR method produces soft, dynamically-generated, multiple probabilistic class labels during successive stages of refinement, which results in significant improvements in the model accuracy. Experiments on three well-known emotion corpora show noticeable gain using the proposed method.

Small adversarial perturbations of input data are able to drastically change performance of machine learning systems, thereby challenging the validity of such systems. We present the very first end-to-end adversarial attacks on a music instrument classification system allowing to add perturbations directly to audio waveforms instead of spectrograms. Our attacks are able to reduce the accuracy close to a random baseline while at the same time keeping perturbations almost imperceptible and producing misclassifications to any desired instrument.

Bengali is a prominent language of the Indian subcontinent. However, while many state-of-the-art acoustic models exist for prominent languages spoken in the region, research and resources for Bengali are few and far between. In this work, we apply CTC based CNN-RNN networks, a prominent deep learning based end-to-end automatic speech recognition technique, to the Bengali ASR task. We also propose and evaluate the applicability and efficacy of small 7x3 and 3x3 convolution kernels which are prominently used in the computer vision domain primarily because of their FLOPs and parameter efficient nature. We propose two CNN blocks, 2-layer Block A and 4-layer Block B, with the first layer comprising of 7x3 kernel and the subsequent layers comprising solely of 3x3 kernels. Using the publicly available Large Bengali ASR Training data set, we benchmark and evaluate the performance of seven deep neural network configurations of varying complexities and depth on the Bengali ASR task. Our best model, with Block B, has a WER of 13.67, having an absolute reduction of 1.39% over comparable model with larger convolution kernels of size 41x11 and 21x11.

Recently, the end-to-end approach has been successfully applied to multi-speaker speech separation and recognition in both single-channel and multichannel conditions. However, severe performance degradation is still observed in the reverberant and noisy scenarios, and there is still a large performance gap between anechoic and reverberant conditions. In this work, we focus on the multichannel multi-speaker reverberant condition, and propose to extend our previous framework for end-to-end dereverberation, beamforming, and speech recognition with improved numerical stability and advanced frontend subnetworks including voice activity detection like masks. The techniques significantly stabilize the end-to-end training process. The experiments on the spatialized wsj1-2mix corpus show that the proposed system achieves about 35% WER relative reduction compared to our conventional multi-channel E2E ASR system, and also obtains decent speech dereverberation and separation performance (SDR=12.5 dB) in the reverberant multi-speaker condition while trained only with the ASR criterion.

In this work, we propose a new approach for language identification using multi-head self-attention combined with raw waveform based 1D convolutional neural networks for Indian languages. Our approach uses an encoder, multi-head selfattention, and a statistics pooling layer. The encoder learns features directly from raw waveforms using 1D convolution kernels and an LSTM layer. The LSTM layer captures temporal information between the features extracted by the 1D convolutional layer. The multi-head self-attention layer takes outputs of the LSTM layer and applies self-attention mechanisms on these features with M different heads. This process helps the model give more weightage to the more useful features and less weightage to the less relevant features. Finally, the frame-level features are combined using a statistics pooling layer to extract the utterance-level feature vector label prediction. We conduct all our experiments on the 373 hrs of audio data for eight different Indian languages. Our experiments show that our approach outperforms the baseline model by an absolute 3.69% improvement in F1-score and achieves the best F1-score of 95.90%. Our approach also shows that using raw waveform models gets a 1.7% improvement in performance compared to the models built using handcrafted features.

Speech is one of the most common forms of communication in humans. Speech commands are essential parts of multimodal controlling of prosthetic hands. In the past decades, researchers used automatic speech recognition systems for controlling prosthetic hands by using speech commands. Automatic speech recognition systems learn how to map human speech to text. Then, they used natural language processing or a look-up table to map the estimated text to a trajectory. However, the performance of conventional speech-controlled prosthetic hands is still unsatisfactory. Recent advancements in general-purpose graphics processing units (GPGPUs) enable intelligent devices to run deep neural networks in real-time. Thus, architectures of intelligent systems have rapidly transformed from the paradigm of composite subsystems optimization to the paradigm of end-to-end optimization. In this paper, we propose an end-to-end convolutional neural network (CNN) that maps speech 2D features directly to trajectories for prosthetic hands. The proposed convolutional neural network is lightweight, and thus it runs in real-time in an embedded GPGPU. The proposed method can use any type of speech 2D feature that has local correlations in each dimension such as spectrogram, MFCC, or PNCC. We omit the speech to text step in controlling the prosthetic hand in this paper. The network is written in Python with Keras library that has a TensorFlow backend. We optimized the CNN for NVIDIA Jetson TX2 developer kit. Our experiment on this CNN demonstrates a root-mean-square error of 0.119 and 20ms running time to produce trajectory outputs corresponding to the voice input data. To achieve a lower error in real-time, we can optimize a similar CNN for a more powerful embedded GPGPU such as NVIDIA AGX Xavier.