In reverberant conditions with a single speaker, each far-field microphone records a reverberant version of the same speaker signal at a different location. In over-determined conditions, where there are more microphones than speakers, each recorded mixture signal can be leveraged as a constraint to narrow down the solutions to target anechoic speech and thereby reduce reverberation. Equipped with this insight, we propose USDnet, a novel deep neural network (DNN) approach for unsupervised speech dereverberation (USD). At each training step, we first feed an input mixture to USDnet to produce an estimate for target speech, and then linearly filter the DNN estimate to approximate the multi-microphone mixture so that the constraint can be satisfied at each microphone, thereby regularizing the DNN estimate to approximate target anechoic speech. The linear filter can be estimated based on the mixture and DNN estimate via neural forward filtering algorithms such as forward convolutive prediction. We show that this novel methodology can promote unsupervised dereverberation of single-source reverberant speech.
Conventional spoofing detection systems have heavily relied on the use of handcrafted features derived from speech data. However, a notable shift has recently emerged towards the direct utilization of raw speech waveforms, as demonstrated by methods like SincNet filters. This shift underscores the demand for more sophisticated audio sample features. Moreover, the success of deep learning models, particularly those utilizing large pretrained wav2vec 2.0 as a featurization front-end, highlights the importance of refined feature encoders. In response, this research assessed the representational capability of wav2vec 2.0 as an audio feature extractor, modifying the size of its pretrained Transformer layers through two key adjustments: (1) selecting a subset of layers starting from the leftmost one and (2) fine-tuning a portion of the selected layers from the rightmost one. We complemented this analysis with five spoofing detection back-end models, with a primary focus on AASIST, enabling us to pinpoint the optimal configuration for the selection and fine-tuning process. In contrast to conventional handcrafted features, our investigation identified several spoofing detection systems that achieve state-of-the-art performance in the ASVspoof 2019 LA dataset. This comprehensive exploration offers valuable insights into feature selection strategies, advancing the field of spoofing detection.
This paper presents a neural vocoder based on a denoising diffusion probabilistic model (DDPM) incorporating explicit periodic signals as auxiliary conditioning signals. Recently, DDPM-based neural vocoders have gained prominence as non-autoregressive models that can generate high-quality waveforms. The neural vocoders based on DDPM have the advantage of training with a simple time-domain loss. In practical applications, such as singing voice synthesis, there is a demand for neural vocoders to generate high-fidelity speech waveforms with flexible pitch control. However, conventional DDPM-based neural vocoders struggle to generate speech waveforms under such conditions. Our proposed model aims to accurately capture the periodic structure of speech waveforms by incorporating explicit periodic signals. Experimental results show that our model improves sound quality and provides better pitch control than conventional DDPM-based neural vocoders.
It is relatively easy to mine a large parallel corpus for any machine learning task, such as speech-to-text or speech-to-speech translation. Although these mined corpora are large in volume, their quality is questionable. This work shows that the simplest filtering technique can trim down these big, noisy datasets to a more manageable, clean dataset. We also show that using this clean dataset can improve the model's performance, as in the case of the multilingual-to-English Speech Translation (ST) model, where, on average, we obtain a 4.65 BLEU score improvement.
Human movement studies and analyses have been fundamental in many scientific domains, ranging from neuroscience to education, pattern recognition to robotics, health care to sports, and beyond. Previous speech motor models were proposed to understand how speech movement is produced and how the resulting speech varies when some parameters are changed. However, the inverse approach, in which the muscular response parameters and the subject's age are derived from real continuous speech, is not possible with such models. Instead, in the handwriting field, the kinematic theory of rapid human movements and its associated Sigma-lognormal model have been applied successfully to obtain the muscular response parameters. This work presents a speech kinematics based model that can be used to study, analyze, and reconstruct complex speech kinematics in a simplified manner. A method based on the kinematic theory of rapid human movements and its associated Sigma lognormal model are applied to describe and to parameterize the asymptotic impulse response of the neuromuscular networks involved in speech as a response to a neuromotor command. The method used to carry out transformations from formants to a movement observation is also presented. Experiments carried out with the (English) VTR TIMIT database and the (German) Saarbrucken Voice Database, including people of different ages, with and without laryngeal pathologies, corroborate the link between the extracted parameters and aging, on the one hand, and the proportion between the first and second formants required in applying the kinematic theory of rapid human movements, on the other. The results should drive innovative developments in the modeling and understanding of speech kinematics.
Diffusion models proved to be powerful models for generative speech enhancement. In recent SGMSE+ approaches, training involves a stochastic differential equation for the diffusion process, adding both Gaussian and environmental noise to the clean speech signal gradually. The speech enhancement performance varies depending on the choice of the stochastic differential equation that controls the evolution of the mean and the variance along the diffusion processes when adding environmental and Gaussian noise. In this work, we highlight that the scale of the variance is a dominant parameter for speech enhancement performance and show that it controls the tradeoff between noise attenuation and speech distortions. More concretely, we show that a larger variance increases the noise attenuation and allows for reducing the computational footprint, as fewer function evaluations for generating the estimate are required.
The rapid progress in personalized speech generation technology, including personalized text-to-speech (TTS) and voice conversion (VC), poses a challenge in distinguishing between generated and real speech for human listeners, resulting in an urgent demand in protecting speakers' voices from malicious misuse. In this regard, we propose a speaker protection method based on adversarial attacks. The proposed method perturbs speech signals by minimally altering the original speech while rendering downstream speech generation models unable to accurately generate the voice of the target speaker. For validation, we employ the open-source pre-trained YourTTS model for speech generation and protect the target speaker's speech in the white-box scenario. Automatic speaker verification (ASV) evaluations were carried out on the generated speech as the assessment of the voice protection capability. Our experimental results show that we successfully perturbed the speaker encoder of the YourTTS model using the gradient-based I-FGSM adversarial perturbation method. Furthermore, the adversarial perturbation is effective in preventing the YourTTS model from generating the speech of the target speaker. Audio samples can be found in https://voiceprivacy.github.io/Adeversarial-Speech-with-YourTTS.
We introduce a new online adaptive filtering method called supervised multi-step adaptive filters (SMS-AF). Our method uses neural networks to control or optimize linear multi-delay or multi-channel frequency-domain filters and can flexibly scale-up performance at the cost of increased compute -- a property rarely addressed in the AF literature, but critical for many applications. To do so, we extend recent work with a set of improvements including feature pruning, a supervised loss, and multiple optimization steps per time-frame. These improvements work in a cohesive manner to unlock scaling. Furthermore, we show how our method relates to Kalman filtering and meta-adaptive filtering, making it seamlessly applicable to a diverse set of AF tasks. We evaluate our method on acoustic echo cancellation (AEC) and multi-channel speech enhancement tasks and compare against several baselines on standard synthetic and real-world datasets. Results show our method performance scales with inference cost and model capacity, yields multi-dB performance gains for both tasks, and is real-time capable on a single CPU core.
Speech intelligibility can be degraded due to multiple factors, such as noisy environments, technical difficulties or biological conditions. This work is focused on the development of an automatic non-intrusive system for predicting the speech intelligibility level in this latter case. The main contribution of our research on this topic is the use of Long Short-Term Memory (LSTM) networks with log-mel spectrograms as input features for this purpose. In addition, this LSTM-based system is further enhanced by the incorporation of a simple attention mechanism that is able to determine the more relevant frames to this task. The proposed models are evaluated with the UA-Speech database that contains dysarthric speech with different degrees of severity. Results show that the attention LSTM architecture outperforms both, a reference Support Vector Machine (SVM)-based system with hand-crafted features and a LSTM-based system with Mean-Pooling.
Despite the widespread adoption, there is a lack of research into how various critical aspects of pretrained language models (PLMs) affect their performance in hate speech detection. Through five research questions, our findings and recommendations lay the groundwork for empirically investigating different aspects of PLMs' use in hate speech detection. We deep dive into comparing different pretrained models, evaluating their seed robustness, finetuning settings, and the impact of pretraining data collection time. Our analysis reveals early peaks for downstream tasks during pretraining, the limited benefit of employing a more recent pretraining corpus, and the significance of specific layers during finetuning. We further call into question the use of domain-specific models and highlight the need for dynamic datasets for benchmarking hate speech detection.