Audio language models have recently emerged as a promising approach for various audio generation tasks, relying on audio tokenizers to encode waveforms into sequences of discrete symbols. Audio tokenization often poses a necessary compromise between code bitrate and reconstruction accuracy. When dealing with low-bitrate audio codes, language models are constrained to process only a subset of the information embedded in the audio, which in turn restricts their generative capabilities. To circumvent these issues, we propose encoding audio as vector sequences in continuous space $\mathbb R^d$ and autoregressively generating these sequences using a decoder-only diffusion transformer (ARDiT). Our findings indicate that ARDiT excels in zero-shot text-to-speech and exhibits performance that compares to or even surpasses that of state-of-the-art models. High-bitrate continuous speech representation enables almost flawless reconstruction, allowing our model to achieve nearly perfect speech editing. Our experiments reveal that employing Integral Kullback-Leibler (IKL) divergence for distillation at each autoregressive step significantly boosts the perceived quality of the samples. Simultaneously, it condenses the iterative sampling process of the diffusion model into a single step. Furthermore, ARDiT can be trained to predict several continuous vectors in one step, significantly reducing latency during sampling. Impressively, one of our models can generate $170$ ms of $24$ kHz speech per evaluation step with minimal degradation in performance. Audio samples are available at http://ardit-tts.github.io/ .
Recent advancements in Latent Diffusion Models (LDMs) have propelled them to the forefront of various generative tasks. However, their iterative sampling process poses a significant computational burden, resulting in slow generation speeds and limiting their application in text-to-audio generation deployment. In this work, we introduce AudioLCM, a novel consistency-based model tailored for efficient and high-quality text-to-audio generation. AudioLCM integrates Consistency Models into the generation process, facilitating rapid inference through a mapping from any point at any time step to the trajectory's initial point. To overcome the convergence issue inherent in LDMs with reduced sample iterations, we propose the Guided Latent Consistency Distillation with a multi-step Ordinary Differential Equation (ODE) solver. This innovation shortens the time schedule from thousands to dozens of steps while maintaining sample quality, thereby achieving fast convergence and high-quality generation. Furthermore, to optimize the performance of transformer-based neural network architectures, we integrate the advanced techniques pioneered by LLaMA into the foundational framework of transformers. This architecture supports stable and efficient training, ensuring robust performance in text-to-audio synthesis. Experimental results on text-to-sound generation and text-to-music synthesis tasks demonstrate that AudioLCM needs only 2 iterations to synthesize high-fidelity audios, while it maintains sample quality competitive with state-of-the-art models using hundreds of steps. AudioLCM enables a sampling speed of 333x faster than real-time on a single NVIDIA 4090Ti GPU, making generative models practically applicable to text-to-audio generation deployment. Our extensive preliminary analysis shows that each design in AudioLCM is effective.
Neural audio synthesis methods now allow specifying ideas in natural language. However, these methods produce results that cannot be easily tweaked, as they are based on large latent spaces and up to billions of uninterpretable parameters. We propose a text-to-audio generation method that leverages a virtual modular sound synthesizer with only 78 parameters. Synthesizers have long been used by skilled sound designers for media like music and film due to their flexibility and intuitive controls. Our method, CTAG, iteratively updates a synthesizer's parameters to produce high-quality audio renderings of text prompts that can be easily inspected and tweaked. Sounds produced this way are also more abstract, capturing essential conceptual features over fine-grained acoustic details, akin to how simple sketches can vividly convey visual concepts. Our results show how CTAG produces sounds that are distinctive, perceived as artistic, and yet similarly identifiable to recent neural audio synthesis models, positioning it as a valuable and complementary tool.
This article presents a review of typical techniques used in three distinct aspects of deep learning model development for audio generation. In the first part of the article, we provide an explanation of audio representations, beginning with the fundamental audio waveform. We then progress to the frequency domain, with an emphasis on the attributes of human hearing, and finally introduce a relatively recent development. The main part of the article focuses on explaining basic and extended deep learning architecture variants, along with their practical applications in the field of audio generation. The following architectures are addressed: 1) Autoencoders 2) Generative adversarial networks 3) Normalizing flows 4) Transformer networks 5) Diffusion models. Lastly, we will examine four distinct evaluation metrics that are commonly employed in audio generation. This article aims to offer novice readers and beginners in the field a comprehensive understanding of the current state of the art in audio generation methods as well as relevant studies that can be explored for future research.
Recent innovations in diffusion probabilistic models have paved the way for significant progress in image, text and audio generation, leading to their applications in generative time series forecasting. However, leveraging such abilities to model highly stochastic time series data remains a challenge. In this paper, we propose a novel Stochastic Diffusion (StochDiff) model which learns data-driven prior knowledge at each time step by utilizing the representational power of the stochastic latent spaces to model the variability of the multivariate time series data. The learnt prior knowledge helps the model to capture complex temporal dynamics and the inherent uncertainty of the data. This improves its ability to model highly stochastic time series data. Through extensive experiments on real-world datasets, we demonstrate the effectiveness of our proposed model on stochastic time series forecasting. Additionally, we showcase an application of our model for real-world surgical guidance, highlighting its potential to benefit the medical community.
While recent advancements in speech language models have achieved significant progress, they face remarkable challenges in modeling the long acoustic sequences of neural audio codecs. In this paper, we introduce \textbf{G}enerative \textbf{P}re-trained \textbf{S}peech \textbf{T}ransformer (GPST), a hierarchical transformer designed for efficient speech language modeling. GPST quantizes audio waveforms into two distinct types of discrete speech representations and integrates them within a hierarchical transformer architecture, allowing for a unified one-stage generation process and enhancing Hi-Res audio generation capabilities. By training on large corpora of speeches in an end-to-end unsupervised manner, GPST can generate syntactically consistent speech with diverse speaker identities. Given a brief 3-second prompt, GPST can produce natural and coherent personalized speech, demonstrating in-context learning abilities. Moreover, our approach can be easily extended to spoken cross-lingual speech generation by incorporating multi-lingual semantic tokens and universal acoustic tokens. Experimental results indicate that GPST significantly outperforms the existing speech language models in terms of word error rate, speech quality, and speaker similarity. The code is available at \url{https://github.com/youngsheen/GPST}.
Video-to-audio (V2A) generation aims to synthesize content-matching audio from silent video, and it remains challenging to build V2A models with high generation quality, efficiency, and visual-audio temporal synchrony. We propose Frieren, a V2A model based on rectified flow matching. Frieren regresses the conditional transport vector field from noise to spectrogram latent with straight paths and conducts sampling by solving ODE, outperforming autoregressive and score-based models in terms of audio quality. By employing a non-autoregressive vector field estimator based on a feed-forward transformer and channel-level cross-modal feature fusion with strong temporal alignment, our model generates audio that is highly synchronized with the input video. Furthermore, through reflow and one-step distillation with guided vector field, our model can generate decent audio in a few, or even only one sampling step. Experiments indicate that Frieren achieves state-of-the-art performance in both generation quality and temporal alignment on VGGSound, with alignment accuracy reaching 97.22%, and 6.2% improvement in inception score over the strong diffusion-based baseline. Audio samples are available at http://frieren-v2a.github.io.
Sound content creation, essential for multimedia works such as video games and films, often involves extensive trial-and-error, enabling creators to semantically reflect their artistic ideas and inspirations, which evolve throughout the creation process, into the sound. Recent high-quality diffusion-based Text-to-Sound (T2S) generative models provide valuable tools for creators. However, these models often suffer from slow inference speeds, imposing an undesirable burden that hinders the trial-and-error process. While existing T2S distillation models address this limitation through 1-step generation, the sample quality of $1$-step generation remains insufficient for production use. Additionally, while multi-step sampling in those distillation models improves sample quality itself, the semantic content changes due to their lack of deterministic sampling capabilities. To address these issues, we introduce Sound Consistency Trajectory Models (SoundCTM), which allow flexible transitions between high-quality $1$-step sound generation and superior sound quality through multi-step deterministic sampling. This allows creators to efficiently conduct trial-and-error with 1-step generation to semantically align samples with their intention, and subsequently refine sample quality with preserving semantic content through deterministic multi-step sampling. To develop SoundCTM, we reframe the CTM training framework, originally proposed in computer vision, and introduce a novel feature distance using the teacher network for a distillation loss. For production-level generation, we scale up our model to 1B trainable parameters, making SoundCTM-DiT-1B the first large-scale distillation model in the sound community to achieve both promising high-quality 1-step and multi-step full-band (44.1kHz) generation.
We introduce C3LLM (Conditioned-on-Three-Modalities Large Language Models), a novel framework combining three tasks of video-to-audio, audio-to-text, and text-to-audio together. C3LLM adapts the Large Language Model (LLM) structure as a bridge for aligning different modalities, synthesizing the given conditional information, and making multimodal generation in a discrete manner. Our contributions are as follows. First, we adapt a hierarchical structure for audio generation tasks with pre-trained audio codebooks. Specifically, we train the LLM to generate audio semantic tokens from the given conditions, and further use a non-autoregressive transformer to generate different levels of acoustic tokens in layers to better enhance the fidelity of the generated audio. Second, based on the intuition that LLMs were originally designed for discrete tasks with the next-word prediction method, we use the discrete representation for audio generation and compress their semantic meanings into acoustic tokens, similar to adding "acoustic vocabulary" to LLM. Third, our method combines the previous tasks of audio understanding, video-to-audio generation, and text-to-audio generation together into one unified model, providing more versatility in an end-to-end fashion. Our C3LLM achieves improved results through various automated evaluation metrics, providing better semantic alignment compared to previous methods.
A white noise signal can access any possible configuration of values, though statistically over many samples tends to a uniform spectral distribution, and is highly unlikely to produce intelligible sound. But how unlikely? The probability that white noise generates a music-like signal over different durations is analyzed, based on some necessary features observed in real music audio signals such as mostly proximate movement and zero crossing rate. Given the mathematical results, the rarity of music as a signal is considered overall. The applicability of this study is not just to show that music has a precious rarity value, but that examination of the size of music relative to the overall size of audio signal space provides information to inform new generations of algorithmic music system (which are now often founded on audio signal generation directly, and may relate to white noise via such machine learning processes as diffusion). Estimated upper bounds on the rarity of music to the size of various physical and musical spaces are compared, to better understand the magnitude of the results (pun intended). Underlying the research are the questions `how much music is still out there?' and `how much music could a machine learning process actually reach?'.
In recent years, with the realistic generation results and a wide range of personalized applications, diffusion-based generative models gain huge attention in both visual and audio generation areas. Compared to the considerable advancements of text2image or text2audio generation, research in audio2visual or visual2audio generation has been relatively slow. The recent audio-visual generation methods usually resort to huge large language model or composable diffusion models. Instead of designing another giant model for audio-visual generation, in this paper we take a step back showing a simple and lightweight generative transformer, which is not fully investigated in multi-modal generation, can achieve excellent results on image2audio generation. The transformer operates in the discrete audio and visual Vector-Quantized GAN space, and is trained in the mask denoising manner. After training, the classifier-free guidance could be deployed off-the-shelf achieving better performance, without any extra training or modification. Since the transformer model is modality symmetrical, it could also be directly deployed for audio2image generation and co-generation. In the experiments, we show that our simple method surpasses recent image2audio generation methods. Generated audio samples can be found at https://docs.google.com/presentation/d/1ZtC0SeblKkut4XJcRaDsSTuCRIXB3ypxmSi7HTY3IyQ/
With the proliferation of Audio Language Model (ALM) based deepfake audio, there is an urgent need for generalized detection methods. ALM-based deepfake audio currently exhibits widespread, high deception, and type versatility, posing a significant challenge to current audio deepfake detection (ADD) models trained solely on vocoded data. To effectively detect ALM-based deepfake audio, we focus on the mechanism of the ALM-based audio generation method, the conversion from neural codec to waveform. We initially constructed the Codecfake dataset, an open-source, large-scale collection comprising over 1 million audio samples in both English and Chinese, focus on ALM-based audio detection. As countermeasure, to achieve universal detection of deepfake audio and tackle domain ascent bias issue of original sharpness aware minimization (SAM), we propose the CSAM strategy to learn a domain balanced and generalized minima. In our experiments, we first demonstrate that ADD model training with the Codecfake dataset can effectively detects ALM-based audio. Furthermore, our proposed generalization countermeasure yields the lowest average equal error rate (EER) of 0.616% across all test conditions compared to baseline models. The dataset and associated code are available online.
In video game design, audio (both environmental background music and object sound effects) play a critical role. Sounds are typically pre-created assets designed for specific locations or objects in a game. However, user-generated content is becoming increasingly popular in modern games (e.g. building custom environments or crafting unique objects). Since the possibilities are virtually limitless, it is impossible for game creators to pre-create audio for user-generated content. We explore the use of generative artificial intelligence to create music and sound effects on-the-fly based on user-generated content. We investigate two avenues for audio generation: 1) text-to-audio: using a text description of user-generated content as input to the audio generator, and 2) image-to-audio: using a rendering of the created environment or object as input to an image-to-text generator, then piping the resulting text description into the audio generator. In this paper we discuss ethical implications of using generative artificial intelligence for user-generated content and highlight two prototype games where audio is generated for user-created environments and objects.
Previous studies on music style transfer have mainly focused on one-to-one style conversion, which is relatively limited. When considering the conversion between multiple styles, previous methods required designing multiple modes to disentangle the complex style of the music, resulting in large computational costs and slow audio generation. The existing music style transfer methods generate spectrograms with artifacts, leading to significant noise in the generated audio. To address these issues, this study proposes a music style transfer framework based on diffusion models (DM) and uses spectrogram-based methods to achieve multi-to-multi music style transfer. The GuideDiff method is used to restore spectrograms to high-fidelity audio, accelerating audio generation speed and reducing noise in the generated audio. Experimental results show that our model has good performance in multi-mode music style transfer compared to the baseline and can generate high-quality audio in real-time on consumer-grade GPUs.
Evolutionary Algorithms and Generative Deep Learning have been two of the most powerful tools for sound generation tasks. However, they have limitations: Evolutionary Algorithms require complicated designs, posing challenges in control and achieving realistic sound generation. Generative Deep Learning models often copy from the dataset and lack creativity. In this paper, we propose LVNS-RAVE, a method to combine Evolutionary Algorithms and Generative Deep Learning to produce realistic and novel sounds. We use the RAVE model as the sound generator and the VGGish model as a novelty evaluator in the Latent Vector Novelty Search (LVNS) algorithm. The reported experiments show that the method can successfully generate diversified, novel audio samples under different mutation setups using different pre-trained RAVE models. The characteristics of the generation process can be easily controlled with the mutation parameters. The proposed algorithm can be a creative tool for sound artists and musicians.
Latent Diffusion Models (LDMs) have emerged as powerful generative models, known for delivering remarkable results under constrained computational resources. However, deploying LDMs on resource-limited devices remains a complex issue, presenting challenges such as memory consumption and inference speed. To address this issue, we introduce LD-Pruner, a novel performance-preserving structured pruning method for compressing LDMs. Traditional pruning methods for deep neural networks are not tailored to the unique characteristics of LDMs, such as the high computational cost of training and the absence of a fast, straightforward and task-agnostic method for evaluating model performance. Our method tackles these challenges by leveraging the latent space during the pruning process, enabling us to effectively quantify the impact of pruning on model performance, independently of the task at hand. This targeted pruning of components with minimal impact on the output allows for faster convergence during training, as the model has less information to re-learn, thereby addressing the high computational cost of training. Consequently, our approach achieves a compressed model that offers improved inference speed and reduced parameter count, while maintaining minimal performance degradation. We demonstrate the effectiveness of our approach on three different tasks: text-to-image (T2I) generation, Unconditional Image Generation (UIG) and Unconditional Audio Generation (UAG). Notably, we reduce the inference time of Stable Diffusion (SD) by 34.9% while simultaneously improving its FID by 5.2% on MS-COCO T2I benchmark. This work paves the way for more efficient pruning methods for LDMs, enhancing their applicability.
Audio-based generative models for music have seen great strides recently, but so far have not managed to produce full-length music tracks with coherent musical structure from text prompts. We show that by training a generative model on long temporal contexts it is possible to produce long-form music of up to 4m45s. Our model consists of a diffusion-transformer operating on a highly downsampled continuous latent representation (latent rate of 21.5Hz). It obtains state-of-the-art generations according to metrics on audio quality and prompt alignment, and subjective tests reveal that it produces full-length music with coherent structure.
Generative multimodal content is increasingly prevalent in much of the content creation arena, as it has the potential to allow artists and media personnel to create pre-production mockups by quickly bringing their ideas to life. The generation of audio from text prompts is an important aspect of such processes in the music and film industry. Many of the recent diffusion-based text-to-audio models focus on training increasingly sophisticated diffusion models on a large set of datasets of prompt-audio pairs. These models do not explicitly focus on the presence of concepts or events and their temporal ordering in the output audio with respect to the input prompt. Our hypothesis is focusing on how these aspects of audio generation could improve audio generation performance in the presence of limited data. As such, in this work, using an existing text-to-audio model Tango, we synthetically create a preference dataset where each prompt has a winner audio output and some loser audio outputs for the diffusion model to learn from. The loser outputs, in theory, have some concepts from the prompt missing or in an incorrect order. We fine-tune the publicly available Tango text-to-audio model using diffusion-DPO (direct preference optimization) loss on our preference dataset and show that it leads to improved audio output over Tango and AudioLDM2, in terms of both automatic- and manual-evaluation metrics.
In recent years, text-to-audio models have revolutionized the field of automatic audio generation. This paper investigates their application in generating synthetic datasets for training data-driven models. Specifically, this study analyzes the performance of two environmental sound classification systems trained with data generated from text-to-audio models. We considered three scenarios: a) augmenting the training dataset with data generated by text-to-audio models; b) using a mixed training dataset combining real and synthetic text-driven generated data; and c) using a training dataset composed entirely of synthetic audio. In all cases, the performance of the classification models was tested on real data. Results indicate that text-to-audio models are effective for dataset augmentation, with consistent performance when replacing a subset of the recorded dataset. However, the performance of the audio recognition models drops when relying entirely on generated audio.
In recent times, the focus on text-to-audio (TTA) generation has intensified, as researchers strive to synthesize audio from textual descriptions. However, most existing methods, though leveraging latent diffusion models to learn the correlation between audio and text embeddings, fall short when it comes to maintaining a seamless synchronization between the produced audio and its video. This often results in discernible audio-visual mismatches. To bridge this gap, we introduce a groundbreaking benchmark for Text-to-Audio generation that aligns with Videos, named T2AV-Bench. This benchmark distinguishes itself with three novel metrics dedicated to evaluating visual alignment and temporal consistency. To complement this, we also present a simple yet effective video-aligned TTA generation model, namely T2AV. Moving beyond traditional methods, T2AV refines the latent diffusion approach by integrating visual-aligned text embeddings as its conditional foundation. It employs a temporal multi-head attention transformer to extract and understand temporal nuances from video data, a feat amplified by our Audio-Visual ControlNet that adeptly merges temporal visual representations with text embeddings. Further enhancing this integration, we weave in a contrastive learning objective, designed to ensure that the visual-aligned text embeddings resonate closely with the audio features. Extensive evaluations on the AudioCaps and T2AV-Bench demonstrate that our T2AV sets a new standard for video-aligned TTA generation in ensuring visual alignment and temporal consistency.