Sound Scene Synthesis at the DCASE 2024 Challenge

This paper presents Task 7 at the DCASE 2024 Challenge: sound scene synthesis. Recent advances in sound synthesis and generative models have enabled the creation of realistic and diverse audio content. We introduce a standardized evaluation framework for comparing different sound scene synthesis systems, incorporating both objective and subjective metrics. The challenge attracted four submissions, which are evaluated using the Fr\'echet Audio Distance (FAD) and human perceptual ratings. Our analysis reveals significant insights into the current capabilities and limitations of sound scene synthesis systems, while also highlighting areas for future improvement in this rapidly evolving field.

Music Discovery Dialogue Generation Using Human Intent Analysis and Large Language Models

A conversational music retrieval system can help users discover music that matches their preferences through dialogue. To achieve this, a conversational music retrieval system should seamlessly engage in multi-turn conversation by 1) understanding user queries and 2) responding with natural language and retrieved music. A straightforward solution would be a data-driven approach utilizing such conversation logs. However, few datasets are available for the research and are limited in terms of volume and quality. In this paper, we present a data generation framework for rich music discovery dialogue using a large language model (LLM) and user intents, system actions, and musical attributes. This is done by i) dialogue intent analysis using grounded theory, ii) generating attribute sequences via cascading database filtering, and iii) generating utterances using large language models. By applying this framework to the Million Song dataset, we create LP-MusicDialog, a Large Language Model based Pseudo Music Dialogue dataset, containing over 288k music conversations using more than 319k music items. Our evaluation shows that the synthetic dataset is competitive with an existing, small human dialogue dataset in terms of dialogue consistency, item relevance, and naturalness. Furthermore, using the dataset, we train a conversational music retrieval model and show promising results.

Challenge on Sound Scene Synthesis: Evaluating Text-to-Audio Generation

Despite significant advancements in neural text-to-audio generation, challenges persist in controllability and evaluation. This paper addresses these issues through the Sound Scene Synthesis challenge held as part of the Detection and Classification of Acoustic Scenes and Events 2024. We present an evaluation protocol combining objective metric, namely Fr\'echet Audio Distance, with perceptual assessments, utilizing a structured prompt format to enable diverse captions and effective evaluation. Our analysis reveals varying performance across sound categories and model architectures, with larger models generally excelling but innovative lightweight approaches also showing promise. The strong correlation between objective metrics and human ratings validates our evaluation approach. We discuss outcomes in terms of audio quality, controllability, and architectural considerations for text-to-audio synthesizers, providing direction for future research.

Correlation of Fréchet Audio Distance With Human Perception of Environmental Audio Is Embedding Dependant

This paper explores whether considering alternative domain-specific embeddings to calculate the Fr\'echet Audio Distance (FAD) metric can help the FAD to correlate better with perceptual ratings of environmental sounds. We used embeddings from VGGish, PANNs, MS-CLAP, L-CLAP, and MERT, which are tailored for either music or environmental sound evaluation. The FAD scores were calculated for sounds from the DCASE 2023 Task 7 dataset. Using perceptual data from the same task, we find that PANNs-WGM-LogMel produces the best correlation between FAD scores and perceptual ratings of both audio quality and perceived fit with a Spearman correlation higher than 0.5. We also find that music-specific embeddings resulted in significantly lower results. Interestingly, VGGish, the embedding used for the original Fr\'echet calculation, yielded a correlation below 0.1. These results underscore the critical importance of the choice of embedding for the FAD metric design.

The Biased Journey of MSD_AUDIO.ZIP

The equitable distribution of academic data is crucial for ensuring equal research opportunities, and ultimately further progress. Yet, due to the complexity of using the API for audio data that corresponds to the Million Song Dataset along with its misreporting (before 2016) and the discontinuation of this API (after 2016), access to this data has become restricted to those within certain affiliations that are connected peer-to-peer. In this paper, we delve into this issue, drawing insights from the experiences of 22 individuals who either attempted to access the data or played a role in its creation. With this, we hope to initiate more critical dialogue and more thoughtful consideration with regard to access privilege in the MIR community.

LP-MusicCaps: LLM-Based Pseudo Music Captioning

Automatic music captioning, which generates natural language descriptions for given music tracks, holds significant potential for enhancing the understanding and organization of large volumes of musical data. Despite its importance, researchers face challenges due to the costly and time-consuming collection process of existing music-language datasets, which are limited in size. To address this data scarcity issue, we propose the use of large language models (LLMs) to artificially generate the description sentences from large-scale tag datasets. This results in approximately 2.2M captions paired with 0.5M audio clips. We term it Large Language Model based Pseudo music caption dataset, shortly, LP-MusicCaps. We conduct a systemic evaluation of the large-scale music captioning dataset with various quantitative evaluation metrics used in the field of natural language processing as well as human evaluation. In addition, we trained a transformer-based music captioning model with the dataset and evaluated it under zero-shot and transfer-learning settings. The results demonstrate that our proposed approach outperforms the supervised baseline model.

HCLAS-X: Hierarchical and Cascaded Lyrics Alignment System Using Multimodal Cross-Correlation

In this work, we address the challenge of lyrics alignment, which involves aligning the lyrics and vocal components of songs. This problem requires the alignment of two distinct modalities, namely text and audio. To overcome this challenge, we propose a model that is trained in a supervised manner, utilizing the cross-correlation matrix of latent representations between vocals and lyrics. Our system is designed in a hierarchical and cascaded manner. It predicts synced time first on a sentence-level and subsequently on a word-level. This design enables the system to process long sequences, as the cross-correlation uses quadratic memory with respect to sequence length. In our experiments, we demonstrate that our proposed system achieves a significant improvement in mean average error, showcasing its robustness in comparison to the previous state-of-the-art model. Additionally, we conduct a qualitative analysis of the system after successfully deploying it in several music streaming services.

A Demand-Driven Perspective on Generative Audio AI

To achieve successful deployment of AI research, it is crucial to understand the demands of the industry. In this paper, we present the results of a survey conducted with professional audio engineers, in order to determine research priorities and define various research tasks. We also summarize the current challenges in audio quality and controllability based on the survey. Our analysis emphasizes that the availability of datasets is currently the main bottleneck for achieving high-quality audio generation. Finally, we suggest potential solutions for some revealed issues with empirical evidence.

Room Impulse Response Estimation in a Multiple Source Environment

In real-world acoustic scenarios, there often are multiple sound sources present in a room. These sources are situated in various locations and produce sounds that reach the listener from multiple directions. The presence of multiple sources in a room creates new challenges in estimating the room impulse response (RIR) as each source has a unique RIR, dependent on its location and orientation. Therefore, issues of determining which RIR should be predicted and how to predict it arise, when the input signal is a mixture of multiple reverberated sources. To address these, we propose a new task of predicting a "representative" RIR for a room in a multiple source environment and present a training method to achieve this goal. In contrast to the model trained in a single source environment, our method shows robust performance, regardless of the number of sources in the environment.

Foley Sound Synthesis at the DCASE 2023 Challenge

The addition of Foley sound effects during post-production is a common technique used to enhance the perceived acoustic properties of multimedia content. Traditionally, Foley sound has been produced by human Foley artists, which involves manual recording and mixing of sound. However, recent advances in sound synthesis and generative models have generated interest in machine-assisted or automatic Foley synthesis techniques. To promote further research in this area, we have organized a challenge in DCASE 2023: Task 7 - Foley Sound Synthesis. Our challenge aims to provide a standardized evaluation framework that is both rigorous and efficient, allowing for the evaluation of different Foley synthesis systems. Through this challenge, we hope to encourage active participation from the research community and advance the state-of-the-art in automatic Foley synthesis. In this technical report, we provide a detailed overview of the Foley sound synthesis challenge, including task definition, dataset, baseline, evaluation scheme and criteria, and discussion.