Music Foundation Model as Generic Booster for Music Downstream Tasks

We demonstrate the efficacy of using intermediate representations from a single foundation model to enhance various music downstream tasks. We introduce SoniDo, a music foundation model (MFM) designed to extract hierarchical features from target music samples. By leveraging hierarchical intermediate features, SoniDo constrains the information granularity, leading to improved performance across various downstream tasks including both understanding and generative tasks. We specifically evaluated this approach on representative tasks such as music tagging, music transcription, music source separation, and music mixing. Our results reveal that the features extracted from foundation models provide valuable enhancements in training downstream task models. This highlights the capability of using features extracted from music foundation models as a booster for downstream tasks. Our approach not only benefits existing task-specific models but also supports music downstream tasks constrained by data scarcity. This paves the way for more effective and accessible music processing solutions.

Mitigating Embedding Collapse in Diffusion Models for Categorical Data

Latent diffusion models have enabled continuous-state diffusion models to handle a variety of datasets, including categorical data. However, most methods rely on fixed pretrained embeddings, limiting the benefits of joint training with the diffusion model. While jointly learning the embedding (via reconstruction loss) and the latent diffusion model (via score matching loss) could enhance performance, our analysis shows that end-to-end training risks embedding collapse, degrading generation quality. To address this issue, we introduce CATDM, a continuous diffusion framework within the embedding space that stabilizes training. We propose a novel objective combining the joint embedding-diffusion variational lower bound with a Consistency-Matching (CM) regularizer, alongside a shifted cosine noise schedule and random dropping strategy. The CM regularizer ensures the recovery of the true data distribution. Experiments on benchmarks show that CATDM mitigates embedding collapse, yielding superior results on FFHQ, LSUN Churches, and LSUN Bedrooms. In particular, CATDM achieves an FID of 6.81 on ImageNet $256\times256$ with 50 steps. It outperforms non-autoregressive models in machine translation and is on a par with previous methods in text generation.

VRVQ: Variable Bitrate Residual Vector Quantization for Audio Compression

Recent state-of-the-art neural audio compression models have progressively adopted residual vector quantization (RVQ). Despite this success, these models employ a fixed number of codebooks per frame, which can be suboptimal in terms of rate-distortion tradeoff, particularly in scenarios with simple input audio, such as silence. To address this limitation, we propose variable bitrate RVQ (VRVQ) for audio codecs, which allows for more efficient coding by adapting the number of codebooks used per frame. Furthermore, we propose a gradient estimation method for the non-differentiable masking operation that transforms from the importance map to the binary importance mask, improving model training via a straight-through estimator. We demonstrate that the proposed training framework achieves superior results compared to the baseline method and shows further improvement when applied to the current state-of-the-art codec.

Distillation of Discrete Diffusion through Dimensional Correlations

Diffusion models have demonstrated exceptional performances in various fields of generative modeling. While they often outperform competitors including VAEs and GANs in sample quality and diversity, they suffer from slow sampling speed due to their iterative nature. Recently, distillation techniques and consistency models are mitigating this issue in continuous domains, but discrete diffusion models have some specific challenges towards faster generation. Most notably, in the current literature, correlations between different dimensions (pixels, locations) are ignored, both by its modeling and loss functions, due to computational limitations. In this paper, we propose "mixture" models in discrete diffusion that are capable of treating dimensional correlations while remaining scalable, and we provide a set of loss functions for distilling the iterations of existing models. Two primary theoretical insights underpin our approach: first, that dimensionally independent models can well approximate the data distribution if they are allowed to conduct many sampling steps, and second, that our loss functions enables mixture models to distill such many-step conventional models into just a few steps by learning the dimensional correlations. We empirically demonstrate that our proposed method for discrete diffusions work in practice, by distilling a continuous-time discrete diffusion model pretrained on the CIFAR-10 dataset.

$\textit{Jump Your Steps}$: Optimizing Sampling Schedule of Discrete Diffusion Models

Diffusion models have seen notable success in continuous domains, leading to the development of discrete diffusion models (DDMs) for discrete variables. Despite recent advances, DDMs face the challenge of slow sampling speeds. While parallel sampling methods like $\tau$-leaping accelerate this process, they introduce $\textit{Compounding Decoding Error}$ (CDE), where discrepancies arise between the true distribution and the approximation from parallel token generation, leading to degraded sample quality. In this work, we present $\textit{Jump Your Steps}$ (JYS), a novel approach that optimizes the allocation of discrete sampling timesteps by minimizing CDE without extra computational cost. More precisely, we derive a practical upper bound on CDE and propose an efficient algorithm for searching for the optimal sampling schedule. Extensive experiments across image, music, and text generation show that JYS significantly improves sampling quality, establishing it as a versatile framework for enhancing DDM performance for fast sampling.

Variable Bitrate Residual Vector Quantization for Audio Coding

Recent state-of-the-art neural audio compression models have progressively adopted residual vector quantization (RVQ). Despite this success, these models employ a fixed number of codebooks per frame, which can be suboptimal in terms of rate-distortion tradeoff, particularly in scenarios with simple input audio, such as silence. To address this limitation, we propose variable bitrate RVQ (VRVQ) for audio codecs, which allows for more efficient coding by adapting the number of codebooks used per frame. Furthermore, we propose a gradient estimation method for the non-differentiable masking operation that transforms from the importance map to the binary importance mask, improving model training via a straight-through estimator. We demonstrate that the proposed training framework achieves superior results compared to the baseline method and shows further improvement when applied to the current state-of-the-art codec.

DisMix: Disentangling Mixtures of Musical Instruments for Source-level Pitch and Timbre Manipulation

Existing work on pitch and timbre disentanglement has been mostly focused on single-instrument music audio, excluding the cases where multiple instruments are presented. To fill the gap, we propose DisMix, a generative framework in which the pitch and timbre representations act as modular building blocks for constructing the melody and instrument of a source, and the collection of which forms a set of per-instrument latent representations underlying the observed mixture. By manipulating the representations, our model samples mixtures with novel combinations of pitch and timbre of the constituent instruments. We can jointly learn the disentangled pitch-timbre representations and a latent diffusion transformer that reconstructs the mixture conditioned on the set of source-level representations. We evaluate the model using both a simple dataset of isolated chords and a realistic four-part chorales in the style of J.S. Bach, identify the key components for the success of disentanglement, and demonstrate the application of mixture transformation based on source-level attribute manipulation.

MoLA: Motion Generation and Editing with Latent Diffusion Enhanced by Adversarial Training

In motion generation, controllability as well as generation quality and speed is becoming more and more important. There are various motion editing tasks, such as in-betweening, upper body editing, and path-following, but existing methods perform motion editing with a data-space diffusion model, which is slow in inference compared to a latent diffusion model. In this paper, we propose MoLA, which provides fast and high-quality motion generation and also can deal with multiple editing tasks in a single framework. For high-quality and fast generation, we employ a variational autoencoder and latent diffusion model, and improve the performance with adversarial training. In addition, we apply a training-free guided generation framework to achieve various editing tasks with motion control inputs. We quantitatively show the effectiveness of adversarial learning in text-to-motion generation, and demonstrate the applicability of our editing framework to multiple editing tasks in the motion domain.

SoundCTM: Uniting Score-based and Consistency Models for Text-to-Sound Generation

Sound content is an indispensable element for multimedia works such as video games, music, and films. Recent high-quality diffusion-based sound generation models can serve as valuable tools for the creators. However, despite producing high-quality sounds, these models often suffer from slow inference speeds. This drawback burdens creators, who typically refine their sounds through trial and error to align them with their artistic intentions. To address this issue, we introduce Sound Consistency Trajectory Models (SoundCTM). Our model enables flexible transitioning between high-quality 1-step sound generation and superior sound quality through multi-step generation. This allows creators to initially control sounds with 1-step samples before refining them through multi-step generation. While CTM fundamentally achieves flexible 1-step and multi-step generation, its impressive performance heavily depends on an additional pretrained feature extractor and an adversarial loss, which are expensive to train and not always available in other domains. Thus, we reframe CTM's training framework and introduce a novel feature distance by utilizing the teacher's network for a distillation loss. Additionally, while distilling classifier-free guided trajectories, we train conditional and unconditional student models simultaneously and interpolate between these models during inference. We also propose training-free controllable frameworks for SoundCTM, leveraging its flexible sampling capability. SoundCTM achieves both promising 1-step and multi-step real-time sound generation without using any extra off-the-shelf networks. Furthermore, we demonstrate SoundCTM's capability of controllable sound generation in a training-free manner.

PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher

To accelerate sampling, diffusion models (DMs) are often distilled into generators that directly map noise to data in a single step. In this approach, the resolution of the generator is fundamentally limited by that of the teacher DM. To overcome this limitation, we propose Progressive Growing of Diffusion Autoencoder (PaGoDA), a technique to progressively grow the resolution of the generator beyond that of the original teacher DM. Our key insight is that a pre-trained, low-resolution DM can be used to deterministically encode high-resolution data to a structured latent space by solving the PF-ODE forward in time (data-to-noise), starting from an appropriately down-sampled image. Using this frozen encoder in an auto-encoder framework, we train a decoder by progressively growing its resolution. From the nature of progressively growing decoder, PaGoDA avoids re-training teacher/student models when we upsample the student model, making the whole training pipeline much cheaper. In experiments, we used our progressively growing decoder to upsample from the pre-trained model's 64x64 resolution to generate 512x512 samples, achieving 2x faster inference compared to single-step distilled Stable Diffusion like LCM. PaGoDA also achieved state-of-the-art FIDs on ImageNet across all resolutions from 64x64 to 512x512. Additionally, we demonstrated PaGoDA's effectiveness in solving inverse problems and enabling controllable generation.