This paper presents the crossing scheme (X-scheme) for improving the performance of deep neural network (DNN)-based music source separation (MSS) without increasing calculation cost. It consists of three components: (i) multi-domain loss (MDL), (ii) bridging operation, which couples the individual instrument networks, and (iii) combination loss (CL). MDL enables the taking advantage of the frequency- and time-domain representations of audio signals. We modify the target network, i.e., the network architecture of the original DNN-based MSS, by adding bridging paths for each output instrument to share their information. MDL is then applied to the combinations of the output sources as well as each independent source, hence we called it CL. MDL and CL can easily be applied to many DNN-based separation methods as they are merely loss functions that are only used during training and do not affect the inference step. Bridging operation does not increase the number of learnable parameters in the network. Experimental results showed that the validity of Open-Unmix (UMX) and densely connected dilated DenseNet (D3Net) extended with our X-scheme, respectively called X-UMX and X-D3Net, by comparing them with their original versions. We also verified the effectiveness of X-scheme in a large-scale data regime, showing its generality with respect to data size. X-UMX Large (X-UMXL), which was trained on large-scale internal data and used in our experiments, is newly available at https://github.com/asteroid-team/asteroid/tree/master/egs/musdb18/X-UMX.
We have developed a diffusion-based speech refiner that improves the reference-free perceptual quality of the audio predicted by preceding single-channel speech separation models. Although modern deep neural network-based speech separation models have show high performance in reference-based metrics, they often produce perceptually unnatural artifacts. The recent advancements made to diffusion models motivated us to tackle this problem by restoring the degraded parts of initial separations with a generative approach. Utilizing the denoising diffusion restoration model (DDRM) as a basis, we propose a shared DDRM-based refiner that generates samples conditioned on the global information of preceding outputs from arbitrary speech separation models. We experimentally show that our refiner can provide a clearer harmonic structure of speech and improves the reference-free metric of perceptual quality for arbitrary preceding model architectures. Furthermore, we tune the variance of the measurement noise based on preceding outputs, which results in higher scores in both reference-free and reference-based metrics. The separation quality can also be further improved by blending the discriminative and generative outputs.
Audio classification and restoration are among major downstream tasks in audio signal processing. However, restoration derives less of a benefit from pretrained models compared to the overwhelming success of pretrained models in classification tasks. Due to such unbalanced benefits, there has been rising interest in how to improve the performance of pretrained models for restoration tasks such as speech enhancement (SE). Previous works have shown that the features extracted by pretrained audio encoders are effective for SE tasks, but these speech-specific encoder-only models usually require extra decoders to become compatible with SE tasks, and involve complicated pretraining procedures or complex data augmentation. Therefore, in pursuit of a universal audio model, the audio masked autoencoder (MAE) whose backbone is the autoencoder of Vision Transformers (ViT-AE), is extended from audio classification toward restoration tasks in this paper. ViT-AE naturally learns mel-to-mel mapping that is compatible with restoration tasks during pretraining. Among many restoration tasks, SE is chosen due to its well-established evaluation metrics and test data. We propose variations of ViT-AE to improve the SE performance, where the mel-to-mel variations yield high scores for non-intrusive metrics and the STFT-oriented variation is effective at standard intrusive metrics such as PESQ. Different variations can be used in accordance with the scenarios. Comprehensive evaluations and ablation studies show that MAE pretraining is also beneficial to SE tasks and help the ViT-AE to better generalize to out-of-domain distortions. We further found that large-scale noisy data of general audio sources, rather than clean speech, is sufficiently effective for pretraining.
Sustaining coherent and engaging narratives requires dialogue or storytelling agents to understand how the personas of speakers or listeners ground the narrative. Specifically, these agents must infer personas of their listeners to produce statements that cater to their interests. They must also learn to maintain consistent speaker personas for themselves throughout the narrative, so that their counterparts feel involved in a realistic conversation or story. However, personas are diverse and complex: they entail large quantities of rich interconnected world knowledge that is challenging to robustly represent in general narrative systems (e.g., a singer is good at singing, and may have attended conservatoire). In this work, we construct a new large-scale persona commonsense knowledge graph, PeaCoK, containing ~100K human-validated persona facts. Our knowledge graph schematizes five dimensions of persona knowledge identified in previous studies of human interactive behaviours, and distils facts in this schema from both existing commonsense knowledge graphs and large-scale pretrained language models. Our analysis indicates that PeaCoK contains rich and precise world persona inferences that help downstream systems generate more consistent and engaging narratives.
Image-to-image translation and voice conversion enable the generation of a new facial image and voice while maintaining some of the semantics such as a pose in an image and linguistic content in audio, respectively. They can aid in the content-creation process in many applications. However, as they are limited to the conversion within each modality, matching the impression of the generated face and voice remains an open question. We propose a cross-modal style transfer framework called XFaVoT that jointly learns four tasks: image translation and voice conversion tasks with audio or image guidance, which enables the generation of ``face that matches given voice" and ``voice that matches given face", and intra-modality translation tasks with a single framework. Experimental results on multiple datasets show that XFaVoT achieves cross-modal style translation of image and voice, outperforming baselines in terms of quality, diversity, and face-voice correspondence.
Although music is typically multi-label, many works have studied hierarchical music tagging with simplified settings such as single-label data. Moreover, there lacks a framework to describe various joint training methods under the multi-label setting. In order to discuss the above topics, we introduce hierarchical multi-label music instrument classification task. The task provides a realistic setting where multi-instrument real music data is assumed. Various hierarchical methods that jointly train a DNN are summarized and explored in the context of the fusion of deep learning and conventional techniques. For the effective joint training in the multi-label setting, we propose two methods to model the connection between fine- and coarse-level tags, where one uses rule-based grouped max-pooling, the other one uses the attention mechanism obtained in a data-driven manner. Our evaluation reveals that the proposed methods have advantages over the method without joint training. In addition, the decision procedure within the proposed methods can be interpreted by visualizing attention maps or referring to fixed rules.
Generative adversarial networks (GANs) learn a target probability distribution by optimizing a generator and a discriminator with minimax objectives. This paper addresses the question of whether such optimization actually provides the generator with gradients that make its distribution close to the target distribution. We derive sufficient conditions for the discriminator to serve as the distance between the distributions by connecting the GAN formulation with the concept of sliced optimal transport. Furthermore, by leveraging these theoretical results, we propose a novel GAN training scheme, called adversarially slicing generative network (ASGN). With only simple modifications, the ASGN is applicable to a broad class of existing GANs. Experiments on synthetic and image datasets support our theoretical results and the ASGN's effectiveness as compared to usual GANs.
Pre-trained diffusion models have been successfully used as priors in a variety of linear inverse problems, where the goal is to reconstruct a signal from noisy linear measurements. However, existing approaches require knowledge of the linear operator. In this paper, we propose GibbsDDRM, an extension of Denoising Diffusion Restoration Models (DDRM) to a blind setting in which the linear measurement operator is unknown. GibbsDDRM constructs a joint distribution of the data, measurements, and linear operator by using a pre-trained diffusion model for the data prior, and it solves the problem by posterior sampling with an efficient variant of a Gibbs sampler. The proposed method is problem-agnostic, meaning that a pre-trained diffusion model can be applied to various inverse problems without fine tuning. In experiments, it achieved high performance on both blind image deblurring and vocal dereverberation tasks, despite the use of simple generic priors for the underlying linear operators.
Recent years have seen progress beyond domain-specific sound separation for speech or music towards universal sound separation for arbitrary sounds. Prior work on universal sound separation has investigated separating a target sound out of an audio mixture given a text query. Such text-queried sound separation systems provide a natural and scalable interface for specifying arbitrary target sounds. However, supervised text-queried sound separation systems require costly labeled audio-text pairs for training. Moreover, the audio provided in existing datasets is often recorded in a controlled environment, causing a considerable generalization gap to noisy audio in the wild. In this work, we aim to approach text-queried universal sound separation by using only unlabeled data. We propose to leverage the visual modality as a bridge to learn the desired audio-textual correspondence. The proposed CLIPSep model first encodes the input query into a query vector using the contrastive language-image pretraining (CLIP) model, and the query vector is then used to condition an audio separation model to separate out the target sound. While the model is trained on image-audio pairs extracted from unlabeled videos, at test time we can instead query the model with text inputs in a zero-shot setting, thanks to the joint language-image embedding learned by the CLIP model. Further, videos in the wild often contain off-screen sounds and background noise that may hinder the model from learning the desired audio-textual correspondence. To address this problem, we further propose an approach called noise invariant training for training a query-based sound separation model on noisy data. Experimental results show that the proposed models successfully learn text-queried universal sound separation using only noisy unlabeled videos, even achieving competitive performance against a supervised model in some settings.
Removing reverb from reverberant music is a necessary technique to clean up audio for downstream music manipulations. Reverberation of music contains two categories, natural reverb, and artificial reverb. Artificial reverb has a wider diversity than natural reverb due to its various parameter setups and reverberation types. However, recent supervised dereverberation methods may fail because they rely on sufficiently diverse and numerous pairs of reverberant observations and retrieved data for training in order to be generalizable to unseen observations during inference. To resolve these problems, we propose an unsupervised method that can remove a general kind of artificial reverb for music without requiring pairs of data for training. The proposed method is based on diffusion models, where it initializes the unknown reverberation operator with a conventional signal processing technique and simultaneously refines the estimate with the help of diffusion models. We show through objective and perceptual evaluations that our method outperforms the current leading vocal dereverberation benchmarks.