The practical aspects of developing an Automatic Speech Recognition System (ASR) with HTK are reviewed. Steps are explained concerning hardware, software, libraries, applications and computer programs used. The common procedure to rapidly apply speech recognition system is summarized. The procedure is illustrated, to implement a speech based electrical switch in home automation for the Indonesian language. The main key of the procedure is to match the environment for training and testing using the training data recorded from the testing program, HVite. Often the silence detector of HTK is wrongly triggered by noises because the microphone is too sensitive. This problem is mitigated by simply scaling down the volume. In this sub-word phone-based speech recognition, noise is included in the training database and labelled particularly. Illustration of the procedure is applied to a home automation application. Electrical switches are controlled by Indonesian speech recognizer. The results show 100% command completion rate.
We investigate densely connected convolutional networks (DenseNets) and their extension with domain adversarial training for noise robust speech recognition. DenseNets are very deep, compact convolutional neural networks which have demonstrated incredible improvements over the state-of-the-art results in computer vision. Our experimental results reveal that DenseNets are more robust against noise than other neural network based models such as deep feed forward neural networks and convolutional neural networks. Moreover, domain adversarial learning can further improve the robustness of DenseNets against both, known and unknown noise conditions.
Speech Emotion Recognition (SER) is one of the essential perceptual methods of humans in understanding the situation and how to interact with others, therefore, in recent years, it has been tried to add the ability to recognize emotions to human-machine communication systems. Since the SER process relies on labeled data, databases are essential for it. Incomplete, low-quality or defective data may lead to inaccurate predictions. In this paper, we fixed the inconsistencies in Sharif Emotional Speech Database (ShEMO), as a Persian database, by using an Automatic Speech Recognition (ASR) system and investigating the effect of Farsi language models obtained from accessible Persian text corpora. We also introduced a Persian/Farsi ASR-based SER system that uses linguistic features of the ASR outputs and Deep Learning-based models.
For many real-world applications, the user-generated inputs usually contain various noises due to speech recognition errors caused by linguistic variations1 or typographical errors (typos). Thus, it is crucial to test model performance on data with realistic input noises to ensure robustness and fairness. However, little study has been done to construct such benchmarks for Chinese, where various language-specific input noises happen in the real world. In order to fill this important gap, we construct READIN: a Chinese multi-task benchmark with REalistic And Diverse Input Noises. READIN contains four diverse tasks and requests annotators to re-enter the original test data with two commonly used Chinese input methods: Pinyin input and speech input. We designed our annotation pipeline to maximize diversity, for example by instructing the annotators to use diverse input method editors (IMEs) for keyboard noises and recruiting speakers from diverse dialectical groups for speech noises. We experiment with a series of strong pretrained language models as well as robust training methods, we find that these models often suffer significant performance drops on READIN even with robustness methods like data augmentation. As the first large-scale attempt in creating a benchmark with noises geared towards user-generated inputs, we believe that READIN serves as an important complement to existing Chinese NLP benchmarks. The source code and dataset can be obtained from https://github.com/thunlp/READIN.
Transformer-based end-to-end speech recognition models have received considerable attention in recent years due to their high training speed and ability to model a long-range global context. Position embedding in the transformer architecture is indispensable because it provides supervision for dependency modeling between elements at different positions in the input sequence. To make use of the time order of the input sequence, many works inject some information about the relative or absolute position of the element into the input sequence. In this work, we investigate various position embedding methods in the convolution-augmented transformer (conformer) and adopt a novel implementation named rotary position embedding (RoPE). RoPE encodes absolute positional information into the input sequence by a rotation matrix, and then naturally incorporates explicit relative position information into a self-attention module. To evaluate the effectiveness of the RoPE method, we conducted experiments on AISHELL-1 and LibriSpeech corpora. Results show that the conformer enhanced with RoPE achieves superior performance in the speech recognition task. Specifically, our model achieves a relative word error rate reduction of 8.70% and 7.27% over the conformer on test-clean and test-other sets of the LibriSpeech corpus respectively.
The goal of speech emotion recognition (SER) is to identify the emotional aspects of speech. The SER challenge for Brazilian Portuguese speech was proposed with short snippets of Portuguese which are classified as neutral, non-neutral female and non-neutral male according to paralinguistic elements (laughing, crying, etc). This dataset contains about $50$ minutes of Brazilian Portuguese speech. As the dataset leans on the small side, we investigate whether a combination of transfer learning and data augmentation techniques can produce positive results. Thus, by combining a data augmentation technique called SpecAugment, with the use of Pretrained Audio Neural Networks (PANNs) for transfer learning we are able to obtain interesting results. The PANNs (CNN6, CNN10 and CNN14) are pretrained on a large dataset called AudioSet containing more than $5000$ hours of audio. They were finetuned on the SER dataset and the best performing model (CNN10) on the validation set was submitted to the challenge, achieving an $F1$ score of $0.73$ up from $0.54$ from the baselines provided by the challenge. Moreover, we also tested the use of Transformer neural architecture, pretrained on about $600$ hours of Brazilian Portuguese audio data. Transformers, as well as more complex models of PANNs (CNN14), fail to generalize to the test set in the SER dataset and do not beat the baseline. Considering the limitation of the dataset sizes, currently the best approach for SER is using PANNs (specifically, CNN6 and CNN10).
In this paper, we propose a novel Siamese Adversarial Network (SAN) architecture for automatic speech recognition, which aims at solving the difficulty of fuzzy audio recognition. Specifically, SAN constructs two sub-networks to differentiate the audio feature input and then introduces a loss to unify the output distribution of these sub-networks. Adversarial learning enables the network to capture more essential acoustic features and helps the models achieve better performance when encountering fuzzy audio input. We conduct numerical experiments with the SAN model on several datasets for the automatic speech recognition task. All experimental results show that the siamese adversarial nets significantly reduce the character error rate (CER). Specifically, we achieve a new state of art 4.37 CER without language model on the AISHELL-1 dataset, which leads to around 5% relative CER reduction. To reveal the generality of the siamese adversarial net, we also conduct experiments on the phoneme recognition task, which also shows the superiority of the siamese adversarial network.
Deep neural networks (DNNs) have achieved excellent results in various tasks, including image and speech recognition. However, optimizing the performance of DNNs requires careful tuning of multiple hyperparameters and network parameters via training. High-performance DNNs utilize a large number of parameters, corresponding to high energy consumption during training. To address these limitations, researchers have developed spiking neural networks (SNNs), which are more energy-efficient and can process data in a biologically plausible manner, making them well-suited for tasks involving sensory data processing, i.e., neuromorphic data. Like DNNs, SNNs are vulnerable to various threats, such as adversarial examples and backdoor attacks. Yet, the attacks and countermeasures for SNNs have been almost fully unexplored. This paper investigates the application of backdoor attacks in SNNs using neuromorphic datasets and different triggers. More precisely, backdoor triggers in neuromorphic data can change their position and color, allowing a larger range of possibilities than common triggers in, e.g., the image domain. We propose different attacks achieving up to 100\% attack success rate without noticeable clean accuracy degradation. We also evaluate the stealthiness of the attacks via the structural similarity metric, showing our most powerful attacks being also stealthy. Finally, we adapt the state-of-the-art defenses from the image domain, demonstrating they are not necessarily effective for neuromorphic data resulting in inaccurate performance.
In this paper we introduce various techniques to improve the performance of electroencephalography (EEG) features based continuous speech recognition (CSR) systems. A connectionist temporal classification (CTC) based automatic speech recognition (ASR) system was implemented for performing recognition. We introduce techniques to initialize the weights of the recurrent layers in the encoder of the CTC model with more meaningful weights rather than with random weights and we make use of an external language model to improve the beam search during decoding time. We finally study the problem of predicting articulatory features from EEG features in this paper.
Many of the recent advances in speech separation are primarily aimed at synthetic mixtures of short audio utterances with high degrees of overlap. These datasets significantly differ from the real conversational data and hence, the models trained and evaluated on these datasets do not generalize to real conversational scenarios. Another issue with using most of these models for long form speech is the nondeterministic ordering of separated speech segments due to either unsupervised clustering for time-frequency masks or Permutation Invariant training (PIT) loss. This leads to difficulty in accurately stitching homogenous speaker segments for downstream tasks like Automatic Speech Recognition (ASR). In this paper, we propose a speaker conditioned separator trained on speaker embeddings extracted directly from the mixed signal. We train this model using a directed loss which regulates the order of the separated segments. With this model, we achieve significant improvements on Word error rate (WER) for real conversational data without the need for an additional re-stitching step.