In this work, we demonstrate that the ptychographic phase problem can be solved in a live fashion during scanning, while data is still being collected. We propose a generally applicable modification of the widespread projection-based algorithms such as Error Reduction (ER) and Difference Map (DM). This novel variant of ptychographic phase retrieval enables immediate visual feedback during experiments, reconstruction of arbitrary-sized objects with a fixed amount of computational resources, and adaptive scanning. By building upon the Real-Time Iterative Spectrogram Inversion (RTISI) family of algorithms from the audio processing literature, we show that live variants of projection-based methods such as DM can be derived naturally and may even achieve higher-quality reconstructions than their classic non-live counterparts with comparable effective computational load.
Much research effort is being applied to the task of compressing the knowledge of self-supervised models, which are powerful, yet large and memory consuming. In this work, we show that the original method of knowledge distillation (and its more recently proposed extension, decoupled knowledge distillation) can be applied to the task of distilling HuBERT. In contrast to methods that focus on distilling internal features, this allows for more freedom in the network architecture of the compressed model. We thus propose to distill HuBERT's Transformer layers into an LSTM-based distilled model that reduces the number of parameters even below DistilHuBERT and at the same time shows improved performance in automatic speech recognition.
Diffusion models have shown promising results in speech enhancement, using a task-adapted diffusion process for the conditional generation of clean speech given a noisy mixture. However, at test time, the neural network used for score estimation is called multiple times to solve the iterative reverse process. This results in a slow inference process and causes discretization errors that accumulate over the sampling trajectory. In this paper, we address these limitations through a two-stage training approach. In the first stage, we train the diffusion model the usual way using the generative denoising score matching loss. In the second stage, we compute the enhanced signal by solving the reverse process and compare the resulting estimate to the clean speech target using a predictive loss. We show that using this second training stage enables achieving the same performance as the baseline model using only 5 function evaluations instead of 60 function evaluations. While the performance of usual generative diffusion algorithms drops dramatically when lowering the number of function evaluations (NFEs) to obtain single-step diffusion, we show that our proposed method keeps a steady performance and therefore largely outperforms the diffusion baseline in this setting and also generalizes better than its predictive counterpart.
Speech emotion conversion is the task of converting the expressed emotion of a spoken utterance to a target emotion while preserving the lexical content and speaker identity. While most existing works in speech emotion conversion rely on acted-out datasets and parallel data samples, in this work we specifically focus on more challenging in-the-wild scenarios and do not rely on parallel data. To this end, we propose a diffusion-based generative model for speech emotion conversion, the EmoConv-Diff, that is trained to reconstruct an input utterance while also conditioning on its emotion. Subsequently, at inference, a target emotion embedding is employed to convert the emotion of the input utterance to the given target emotion. As opposed to performing emotion conversion on categorical representations, we use a continuous arousal dimension to represent emotions while also achieving intensity control. We validate the proposed methodology on a large in-the-wild dataset, the MSP-Podcast v1.10. Our results show that the proposed diffusion model is indeed capable of synthesizing speech with a controllable target emotion. Crucially, the proposed approach shows improved performance along the extreme values of arousal and thereby addresses a common challenge in the speech emotion conversion literature.
Several recent contributions in the field of iterative STFT phase retrieval have demonstrated that the performance of the classical Griffin-Lim method can be considerably improved upon. By using the same projection operators as Griffin-Lim, but combining them in innovative ways, these approaches achieve better results in terms of both reconstruction quality and required number of iterations, while retaining a similar computational complexity per iteration. However, like Griffin-Lim, these algorithms operate in an offline manner and thus require an entire spectrogram as input, which is an unrealistic requirement for many real-world speech communication applications. We propose to extend RTISI -- an existing online (frame-by-frame) variant of the Griffin-Lim algorithm -- into a flexible framework that enables straightforward online implementation of any algorithm based on iterative projections. We further employ this framework to implement online variants of the fast Griffin-Lim algorithm, the accelerated Griffin-Lim algorithm, and two algorithms from the optics domain. Evaluation results on speech signals show that, similarly to the offline case, these algorithms can achieve a considerable performance gain compared to RTISI.
In this paper we present a method for single-channel wind noise reduction using our previously proposed diffusion-based stochastic regeneration model combining predictive and generative modelling. We introduce a non-additive speech in noise model to account for the non-linear deformation of the membrane caused by the wind flow and possible clipping. We show that our stochastic regeneration model outperforms other neural-network-based wind noise reduction methods as well as purely predictive and generative models, on a dataset using simulated and real-recorded wind noise. We further show that the proposed method generalizes well by testing on an unseen dataset with real-recorded wind noise. Audio samples, data generation scripts and code for the proposed methods can be found online (https://uhh.de/inf-sp-storm-wind).
We present in this paper an informed single-channel dereverberation method based on conditional generation with diffusion models. With knowledge of the room impulse response, the anechoic utterance is generated via reverse diffusion using a measurement consistency criterion coupled with a neural network that represents the clean speech prior. The proposed approach is largely more robust to measurement noise compared to a state-of-the-art informed single-channel dereverberation method, especially for non-stationary noise. Furthermore, we compare to other blind dereverberation methods using diffusion models and show superiority of the proposed approach for large reverberation times. We motivate the presented algorithm by introducing an extension for blind dereverberation allowing joint estimation of the room impulse response and anechoic speech. Audio samples and code can be found online (https://uhh.de/inf-sp-derev-dps).
Since its inception, the field of deep speech enhancement has been dominated by predictive (discriminative) approaches, such as spectral mapping or masking. Recently, however, novel generative approaches have been applied to speech enhancement, attaining good denoising performance with high subjective quality scores. At the same time, advances in deep learning also allowed for the creation of neural network-based metrics, which have desirable traits such as being able to work without a reference (non-intrusively). Since generatively enhanced speech tends to exhibit radically different residual distortions, its evaluation using instrumental speech metrics may behave differently compared to predictively enhanced speech. In this paper, we evaluate the performance of the same speech enhancement backbone trained under predictive and generative paradigms on a variety of metrics and show that intrusive and non-intrusive measures correlate differently for each paradigm. This analysis motivates the search for metrics that can together paint a complete and unbiased picture of speech enhancement performance, irrespective of the model's training process.
Speech emotion conversion aims to convert the expressed emotion of a spoken utterance to a target emotion while preserving the lexical information and the speaker's identity. In this work, we specifically focus on in-the-wild emotion conversion where parallel data does not exist, and the problem of disentangling lexical, speaker, and emotion information arises. In this paper, we introduce a methodology that uses self-supervised networks to disentangle the lexical, speaker, and emotional content of the utterance, and subsequently uses a HiFiGAN vocoder to resynthesise the disentangled representations to a speech signal of the targeted emotion. For better representation and to achieve emotion intensity control, we specifically focus on the aro\-usal dimension of continuous representations, as opposed to performing emotion conversion on categorical representations. We test our methodology on the large in-the-wild MSP-Podcast dataset. Results reveal that the proposed approach is aptly conditioned on the emotional content of input speech and is capable of synthesising natural-sounding speech for a target emotion. Results further reveal that the methodology better synthesises speech for mid-scale arousal (2 to 6) than for extreme arousal (1 and 7).
This paper introduces an audio-visual speech enhancement system that leverages score-based generative models, also known as diffusion models, conditioned on visual information. In particular, we exploit audio-visual embeddings obtained from a self-super\-vised learning model that has been fine-tuned on lipreading. The layer-wise features of its transformer-based encoder are aggregated, time-aligned, and incorporated into the noise conditional score network. Experimental evaluations show that the proposed audio-visual speech enhancement system yields improved speech quality and reduces generative artifacts such as phonetic confusions with respect to the audio-only equivalent. The latter is supported by the word error rate of a downstream automatic speech recognition model, which decreases noticeably, especially at low input signal-to-noise ratios.