From Aesthetics to Human Preferences: Comparative Perspectives of Evaluating Text-to-Music Systems

Evaluating generative models remains a fundamental challenge, particularly when the goal is to reflect human preferences. In this paper, we use music generation as a case study to investigate the gap between automatic evaluation metrics and human preferences. We conduct comparative experiments across five state-of-the-art music generation approaches, assessing both perceptual quality and distributional similarity to human-composed music. Specifically, we evaluate synthesis music from various perceptual dimensions and examine reference-based metrics such as Mauve Audio Divergence (MAD) and Kernel Audio Distance (KAD). Our findings reveal significant inconsistencies across the different metrics, highlighting the limitation of the current evaluation practice. To support further research, we release a benchmark dataset comprising samples from multiple models. This study provides a broader perspective on the alignment of human preference in generative modeling, advocating for more human-centered evaluation strategies across domains.

JaccDiv: A Metric and Benchmark for Quantifying Diversity of Generated Marketing Text in the Music Industry

Online platforms are increasingly interested in using Data-to-Text technologies to generate content and help their users. Unfortunately, traditional generative methods often fall into repetitive patterns, resulting in monotonous galleries of texts after only a few iterations. In this paper, we investigate LLM-based data-to-text approaches to automatically generate marketing texts that are of sufficient quality and diverse enough for broad adoption. We leverage Language Models such as T5, GPT-3.5, GPT-4, and LLaMa2 in conjunction with fine-tuning, few-shot, and zero-shot approaches to set a baseline for diverse marketing texts. We also introduce a metric JaccDiv to evaluate the diversity of a set of texts. This research extends its relevance beyond the music industry, proving beneficial in various fields where repetitive automated content generation is prevalent.

Solving Copyright Infringement on Short Video Platforms: Novel Datasets and an Audio Restoration Deep Learning Pipeline

Short video platforms like YouTube Shorts and TikTok face significant copyright compliance challenges, as infringers frequently embed arbitrary background music (BGM) to obscure original soundtracks (OST) and evade content originality detection. To tackle this issue, we propose a novel pipeline that integrates Music Source Separation (MSS) and cross-modal video-music retrieval (CMVMR). Our approach effectively separates arbitrary BGM from the original OST, enabling the restoration of authentic video audio tracks. To support this work, we introduce two domain-specific datasets: OASD-20K for audio separation and OSVAR-160 for pipeline evaluation. OASD-20K contains 20,000 audio clips featuring mixed BGM and OST pairs, while OSVAR160 is a unique benchmark dataset comprising 1,121 video and mixed-audio pairs, specifically designed for short video restoration tasks. Experimental results demonstrate that our pipeline not only removes arbitrary BGM with high accuracy but also restores OSTs, ensuring content integrity. This approach provides an ethical and scalable solution to copyright challenges in user-generated content on short video platforms.

MusFlow: Multimodal Music Generation via Conditional Flow Matching

Music generation aims to create music segments that align with human aesthetics based on diverse conditional information. Despite advancements in generating music from specific textual descriptions (e.g., style, genre, instruments), the practical application is still hindered by ordinary users' limited expertise or time to write accurate prompts. To bridge this application gap, this paper introduces MusFlow, a novel multimodal music generation model using Conditional Flow Matching. We employ multiple Multi-Layer Perceptrons (MLPs) to align multimodal conditional information into the audio's CLAP embedding space. Conditional flow matching is trained to reconstruct the compressed Mel-spectrogram in the pretrained VAE latent space guided by aligned feature embedding. MusFlow can generate music from images, story texts, and music captions. To collect data for model training, inspired by multi-agent collaboration, we construct an intelligent data annotation workflow centered around a fine-tuned Qwen2-VL model. Using this workflow, we build a new multimodal music dataset, MMusSet, with each sample containing a quadruple of image, story text, music caption, and music piece. We conduct four sets of experiments: image-to-music, story-to-music, caption-to-music, and multimodal music generation. Experimental results demonstrate that MusFlow can generate high-quality music pieces whether the input conditions are unimodal or multimodal. We hope this work can advance the application of music generation in multimedia field, making music creation more accessible. Our generated samples, code and dataset are available at musflow.github.io.

Apollo: An Interactive Environment for Generating Symbolic Musical Phrases using Corpus-based Style Imitation

With the recent developments in machine intelligence and web technologies, new generative music systems are being explored for assisted composition using machine learning techniques on the web. Such systems are built for various tasks such as melodic, harmonic or rhythm generation, music interpolation, continuation and style imitation. In this paper, we introduce Apollo, an interactive music application for generating symbolic phrases of conventional western music using corpus-based style imitation techniques. In addition to enabling the construction and management of symbolic musical corpora, the system makes it possible for music artists and researchers to generate new musical phrases in the style of the proposed corpus. The system is available as a desktop application. The generated symbolic music materials, encoded in the MIDI format, can be exported or streamed for various purposes including using them as seed material for musical projects. We present the system design, implementation details, discuss and conclude with future work for the system.

Calliope: An Online Generative Music System for Symbolic Multi-Track Composition

With the rise of artificial intelligence in recent years, there has been a rapid increase in its application towards creative domains, including music. There exist many systems built that apply machine learning approaches to the problem of computer-assisted music composition (CAC). Calliope is a web application that assists users in performing a variety of multi-track composition tasks in the symbolic domain. The user can upload (Musical Instrument Digital Interface) MIDI files, visualize and edit MIDI tracks, and generate partial (via bar in-filling) or complete multi-track content using the Multi-Track Music Machine (MMM). Generation of new MIDI excerpts can be done in batch and can be combined with active playback listening for an enhanced assisted-composition workflow. The user can export generated MIDI materials or directly stream MIDI playback from the system to their favorite Digital Audio Workstation (DAW). We present a demonstration of the system, its features, generative parameters and describe the co-creative workflows that it affords.

Kimi-Audio Technical Report

We present Kimi-Audio, an open-source audio foundation model that excels in audio understanding, generation, and conversation. We detail the practices in building Kimi-Audio, including model architecture, data curation, training recipe, inference deployment, and evaluation. Specifically, we leverage a 12.5Hz audio tokenizer, design a novel LLM-based architecture with continuous features as input and discrete tokens as output, and develop a chunk-wise streaming detokenizer based on flow matching. We curate a pre-training dataset that consists of more than 13 million hours of audio data covering a wide range of modalities including speech, sound, and music, and build a pipeline to construct high-quality and diverse post-training data. Initialized from a pre-trained LLM, Kimi-Audio is continual pre-trained on both audio and text data with several carefully designed tasks, and then fine-tuned to support a diverse of audio-related tasks. Extensive evaluation shows that Kimi-Audio achieves state-of-the-art performance on a range of audio benchmarks including speech recognition, audio understanding, audio question answering, and speech conversation. We release the codes, model checkpoints, as well as the evaluation toolkits in https://github.com/MoonshotAI/Kimi-Audio.

Evaluating Human-AI Interaction via Usability, User Experience and Acceptance Measures for MMM-C: A Creative AI System for Music Composition

With the rise of artificial intelligence (AI), there has been increasing interest in human-AI co-creation in a variety of artistic domains including music as AI-driven systems are frequently able to generate human-competitive artifacts. Now, the implications of such systems for musical practice are being investigated. We report on a thorough evaluation of the user adoption of the Multi-Track Music Machine (MMM) as a co-creative AI tool for music composers. To do this, we integrate MMM into Cubase, a popular Digital Audio Workstation (DAW) by Steinberg, by producing a "1-parameter" plugin interface named MMM-Cubase (MMM-C), which enables human-AI co-composition. We contribute a methodological assemblage as a 3-part mixed method study measuring usability, user experience and technology acceptance of the system across two groups of expert-level composers: hobbyists and professionals. Results show positive usability and acceptance scores. Users report experiences of novelty, surprise and ease of use from using the system, and limitations on controllability and predictability of the interface when generating music. Findings indicate no significant difference between the two user groups.

DRAGON: Distributional Rewards Optimize Diffusion Generative Models

We present Distributional RewArds for Generative OptimizatioN (DRAGON), a versatile framework for fine-tuning media generation models towards a desired outcome. Compared with traditional reinforcement learning with human feedback (RLHF) or pairwise preference approaches such as direct preference optimization (DPO), DRAGON is more flexible. It can optimize reward functions that evaluate either individual examples or distributions of them, making it compatible with a broad spectrum of instance-wise, instance-to-distribution, and distribution-to-distribution rewards. Leveraging this versatility, we construct novel reward functions by selecting an encoder and a set of reference examples to create an exemplar distribution. When cross-modality encoders such as CLAP are used, the reference examples may be of a different modality (e.g., text versus audio). Then, DRAGON gathers online and on-policy generations, scores them to construct a positive demonstration set and a negative set, and leverages the contrast between the two sets to maximize the reward. For evaluation, we fine-tune an audio-domain text-to-music diffusion model with 20 different reward functions, including a custom music aesthetics model, CLAP score, Vendi diversity, and Frechet audio distance (FAD). We further compare instance-wise (per-song) and full-dataset FAD settings while ablating multiple FAD encoders and reference sets. Over all 20 target rewards, DRAGON achieves an 81.45% average win rate. Moreover, reward functions based on exemplar sets indeed enhance generations and are comparable to model-based rewards. With an appropriate exemplar set, DRAGON achieves a 60.95% human-voted music quality win rate without training on human preference annotations. As such, DRAGON exhibits a new approach to designing and optimizing reward functions for improving human-perceived quality. Sound examples at https://ml-dragon.github.io/web.

Generation of Musical Timbres using a Text-Guided Diffusion Model

In recent years, text-to-audio systems have achieved remarkable success, enabling the generation of complete audio segments directly from text descriptions. While these systems also facilitate music creation, the element of human creativity and deliberate expression is often limited. In contrast, the present work allows composers, arrangers, and performers to create the basic building blocks for music creation: audio of individual musical notes for use in electronic instruments and DAWs. Through text prompts, the user can specify the timbre characteristics of the audio. We introduce a system that combines a latent diffusion model and multi-modal contrastive learning to generate musical timbres conditioned on text descriptions. By jointly generating the magnitude and phase of the spectrogram, our method eliminates the need for subsequently running a phase retrieval algorithm, as related methods do. Audio examples, source code, and a web app are available at https://wxuanyuan.github.io/Musical-Note-Generation/