The generation of high-quality medical time series data is essential for advancing healthcare diagnostics and safeguarding patient privacy. Specifically, synthesizing realistic phonocardiogram (PCG) signals offers significant potential as a cost-effective and efficient tool for cardiac disease pre-screening. Despite its potential, the synthesis of PCG signals for this specific application received limited attention in research. In this study, we employ and compare three state-of-the-art generative models from different categories - WaveNet, DoppelGANger, and DiffWave - to generate high-quality PCG data. We use data from the George B. Moody PhysioNet Challenge 2022. Our methods are evaluated using various metrics widely used in the previous literature in the domain of time series data generation, such as mean absolute error and maximum mean discrepancy. Our results demonstrate that the generated PCG data closely resembles the original datasets, indicating the effectiveness of our generative models in producing realistic synthetic PCG data. In our future work, we plan to incorporate this method into a data augmentation pipeline to synthesize abnormal PCG signals with heart murmurs, in order to address the current scarcity of abnormal data. We hope to improve the robustness and accuracy of diagnostic tools in cardiology, enhancing their effectiveness in detecting heart murmurs.

This paper introduces UnDiff, a diffusion probabilistic model capable of solving various speech inverse tasks. Being once trained for speech waveform generation in an unconditional manner, it can be adapted to different tasks including degradation inversion, neural vocoding, and source separation. In this paper, we, first, tackle the challenging problem of unconditional waveform generation by comparing different neural architectures and preconditioning domains. After that, we demonstrate how the trained unconditional diffusion could be adapted to different tasks of speech processing by the means of recent developments in post-training conditioning of diffusion models. Finally, we demonstrate the performance of the proposed technique on the tasks of bandwidth extension, declipping, vocoding, and speech source separation and compare it to the baselines. The codes are publicly available.

Recent years have seen considerable advances in audio synthesis with deep generative models. However, the state-of-the-art is very difficult to quantify; different studies often use different evaluation methodologies and different metrics when reporting results, making a direct comparison to other systems difficult if not impossible. Furthermore, the perceptual relevance and meaning of the reported metrics in most cases unknown, prohibiting any conclusive insights with respect to practical usability and audio quality. This paper presents a study that investigates state-of-the-art approaches side-by-side with (i) a set of previously proposed objective metrics for audio reconstruction, and with (ii) a listening study. The results indicate that currently used objective metrics are insufficient to describe the perceptual quality of current systems.

In this work, we propose DiffWave, a versatile diffusion probabilistic model for conditional and unconditional waveform generation. The model is non-autoregressive, and converts the white noise signal into structured waveform through a Markov chain with a constant number of steps at synthesis. It is efficiently trained by optimizing a variant of variational bound on the data likelihood. DiffWave produces high-fidelity audio in different waveform generation tasks, including neural vocoding conditioned on mel spectrogram, class-conditional generation, and unconditional generation. We demonstrate that DiffWave matches a strong WaveNet vocoder in terms of speech quality (MOS: 4.44 versus 4.43), while synthesizing orders of magnitude faster. In particular, it significantly outperforms autoregressive and GAN-based waveform models in the challenging unconditional generation task in terms of audio quality and sample diversity from various automatic and human evaluations. Deep generative models have produced high-fidelity raw audio in speech synthesis and music generation. In previous work, likelihood-based models, including autoregressive models (van den Oord et al., 2016; Kalchbrenner et al., 2018; Mehri et al., 2017) and flow-based models (Prenger et al., 2019; Ping et al., 2020; Kim et al., 2019), have predominated in audio synthesis because of the simple training objective and superior ability of modeling the fine details of waveform in real data.