We investigate the effectiveness of generative adversarial networks (GANs) for speech enhancement, in the context of improving noise robustness of automatic speech recognition (ASR) systems. Prior work demonstrates that GANs can effectively suppress additive noise in raw waveform speech signals, improving perceptual quality metrics; however this technique was not justified in the context of ASR. In this work, we conduct a detailed study to measure the effectiveness of GANs in enhancing speech contaminated by both additive and reverberant noise. Motivated by recent advances in image processing, we propose operating GANs on log-Mel filterbank spectra instead of waveforms, which requires less computation and is more robust to reverberant noise. While GAN enhancement improves the performance of a clean-trained ASR system on noisy speech, it falls short of the performance achieved by conventional multi-style training (MTR). By appending the GAN-enhanced features to the noisy inputs and retraining, we achieve a 7% WER improvement relative to the MTR system.
Automatic speech recognition (ASR) systems often need to be developed for extremely low-resource languages to serve end-uses such as audio content categorization and search. While universal phone recognition is natural to consider when no transcribed speech is available to train an ASR system in a language, adapting universal phone models using very small amounts (minutes rather than hours) of transcribed speech also needs to be studied, particularly with state-of-the-art DNN-based acoustic models. The DARPA LORELEI program provides a framework for such very-low-resource ASR studies, and provides an extrinsic metric for evaluating ASR performance in a humanitarian assistance, disaster relief setting. This paper presents our Kaldi-based systems for the program, which employ a universal phone modeling approach to ASR, and describes recipes for very rapid adaptation of this universal ASR system. The results we obtain significantly outperform results obtained by many competing approaches on the NIST LoReHLT 2017 Evaluation datasets.
Creating universal speaker encoders which are robust for different acoustic and speech duration conditions is a big challenge today. According to our observations systems trained on short speech segments are optimal for short phrase speaker verification and systems trained on long segments are superior for long segments verification. A system trained simultaneously on pooled short and long speech segments does not give optimal verification results and usually degrades both for short and long segments. This paper addresses the problem of creating universal speaker encoders for different speech segments duration. We describe our simple recipe for training universal speaker encoder for any type of selected neural network architecture. According to our evaluation results of wav2vec-TDNN based systems obtained for NIST SRE and VoxCeleb1 benchmarks the proposed universal encoder provides speaker verification improvements in case of different enrollment and test speech segment duration. The key feature of the proposed encoder is that it has the same inference time as the selected neural network architecture.
Automatic speech emotion recognition (SER) is a challenging task that plays a crucial role in natural human-computer interaction. One of the main challenges in SER is data scarcity, i.e., insufficient amounts of carefully labeled data to build and fully explore complex deep learning models for emotion classification. This paper aims to address this challenge using a transfer learning strategy combined with spectrogram augmentation. Specifically, we propose a transfer learning approach that leverages a pre-trained residual network (ResNet) model including a statistics pooling layer from speaker recognition trained using large amounts of speaker-labeled data. The statistics pooling layer enables the model to efficiently process variable-length input, thereby eliminating the need for sequence truncation which is commonly used in SER systems. In addition, we adopt a spectrogram augmentation technique to generate additional training data samples by applying random time-frequency masks to log-mel spectrograms to mitigate overfitting and improve the generalization of emotion recognition models. We evaluate the effectiveness of our proposed approach on the interactive emotional dyadic motion capture (IEMOCAP) dataset. Experimental results indicate that the transfer learning and spectrogram augmentation approaches improve the SER performance, and when combined achieve state-of-the-art results.
Noise robustness is essential for deploying automatic speech recognition (ASR) systems in real-world environments. One way to reduce the effect of noise interference is to employ a preprocessing module that conducts speech enhancement, and then feed the enhanced speech to an ASR backend. In this work, instead of suppressing background noise with a conventional cascaded pipeline, we employ a noise-robust representation learned by a refined self-supervised framework for noisy speech recognition. We propose to combine a reconstruction module with contrastive learning and perform multi-task continual pre-training on noisy data. The reconstruction module is used for auxiliary learning to improve the noise robustness of the learned representation and thus is not required during inference. Experiments demonstrate the effectiveness of our proposed method. Our model substantially reduces the word error rate (WER) for the synthesized noisy LibriSpeech test sets, and yields around 4.1/7.5% WER reduction on noisy clean/other test sets compared to data augmentation. For the real-world noisy speech from the CHiME-4 challenge (1-channel track), we have obtained the state of the art ASR performance without any denoising front-end. Moreover, we achieve comparable performance to the best supervised approach reported with only 16% of labeled data.
In this manuscript, the topic of multi-corpus Speech Emotion Recognition (SER) is approached from a deep transfer learning perspective. A large corpus of emotional speech data, EmoSet, is assembled from a number of existing SER corpora. In total, EmoSet contains 84181 audio recordings from 26 SER corpora with a total duration of over 65 hours. The corpus is then utilised to create a novel framework for multi-corpus speech emotion recognition, namely EmoNet. A combination of a deep ResNet architecture and residual adapters is transferred from the field of multi-domain visual recognition to multi-corpus SER on EmoSet. Compared against two suitable baselines and more traditional training and transfer settings for the ResNet, the residual adapter approach enables parameter efficient training of a multi-domain SER model on all 26 corpora. A shared model with only $3.5$ times the number of parameters of a model trained on a single database leads to increased performance for 21 of the 26 corpora in EmoSet. Measured by McNemar's test, these improvements are further significant for ten datasets at $p<0.05$ while there are just two corpora that see only significant decreases across the residual adapter transfer experiments. Finally, we make our EmoNet framework publicly available for users and developers at https://github.com/EIHW/EmoNet. EmoNet provides an extensive command line interface which is comprehensively documented and can be used in a variety of multi-corpus transfer learning settings.
In this paper, we propose an online attention mechanism, known as cumulative attention (CA), for streaming Transformer-based automatic speech recognition (ASR). Inspired by monotonic chunkwise attention (MoChA) and head-synchronous decoder-end adaptive computation steps (HS-DACS) algorithms, CA triggers the ASR outputs based on the acoustic information accumulated at each encoding timestep, where the decisions are made using a trainable device, referred to as halting selector. In CA, all the attention heads of the same decoder layer are synchronised to have a unified halting position. This feature effectively alleviates the problem caused by the distinct behaviour of individual heads, which may otherwise give rise to severe latency issues as encountered by MoChA. The ASR experiments conducted on AIShell-1 and Librispeech datasets demonstrate that the proposed CA-based Transformer system can achieve on par or better performance with significant reduction in latency during inference, when compared to other streaming Transformer systems in literature.
In recent years, end-to-end (E2E) based automatic speech recognition (ASR) systems have achieved great success due to their simplicity and promising performance. Neural Transducer based models are increasingly popular in streaming E2E based ASR systems and have been reported to outperform the traditional hybrid system in some scenarios. However, the joint optimization of acoustic model, lexicon and language model in neural Transducer also brings about challenges to utilize pure text for language model adaptation. This drawback might prevent their potential applications in practice. In order to address this issue, in this paper, we propose a novel model, factorized neural Transducer, by factorizing the blank and vocabulary prediction, and adopting a standalone language model for the vocabulary prediction. It is expected that this factorization can transfer the improvement of the standalone language model to the Transducer for speech recognition, which allows various language model adaptation techniques to be applied. We demonstrate that the proposed factorized neural Transducer yields 15% to 20% WER improvements when out-of-domain text data is used for language model adaptation, at the cost of a minor degradation in WER on a general test set.
In this paper, we propose a neural end-to-end system for voice preserving, lip-synchronous translation of videos. The system is designed to combine multiple component models and produces a video of the original speaker speaking in the target language that is lip-synchronous with the target speech, yet maintains emphases in speech, voice characteristics, face video of the original speaker. The pipeline starts with automatic speech recognition including emphasis detection, followed by a translation model. The translated text is then synthesized by a Text-to-Speech model that recreates the original emphases mapped from the original sentence. The resulting synthetic voice is then mapped back to the original speakers' voice using a voice conversion model. Finally, to synchronize the lips of the speaker with the translated audio, a conditional generative adversarial network-based model generates frames of adapted lip movements with respect to the input face image as well as the output of the voice conversion model. In the end, the system combines the generated video with the converted audio to produce the final output. The result is a video of a speaker speaking in another language without actually knowing it. To evaluate our design, we present a user study of the complete system as well as separate evaluations of the single components. Since there is no available dataset to evaluate our whole system, we collect a test set and evaluate our system on this test set. The results indicate that our system is able to generate convincing videos of the original speaker speaking the target language while preserving the original speaker's characteristics. The collected dataset will be shared.
A language agnostic approach to recognizing emotions from speech remains an incomplete and challenging task. In this paper, we used Bangla and English languages to assess whether distinguishing emotions from speech is independent of language. The following emotions were categorized for this study: happiness, anger, neutral, sadness, disgust, and fear. We employed three Emotional Speech Sets, of which the first two were developed by native Bengali speakers in Bangla and English languages separately. The third was the Toronto Emotional Speech Set (TESS), which was developed by native English speakers from Canada. We carefully selected language-independent prosodic features, adopted a Support Vector Machine (SVM) model, and conducted three experiments to carry out our proposition. In the first experiment, we measured the performance of the three speech sets individually. This was followed by the second experiment, where we recorded the classification rate by combining the speech sets. Finally, in the third experiment we measured the recognition rate by training and testing the model with different speech sets. Although this study reveals that Speech Emotion Recognition (SER) is mostly language-independent, there is some disparity while recognizing emotional states like disgust and fear in these two languages. Moreover, our investigations inferred that non-native speakers convey emotions through speech, much like expressing themselves in their native tongue.