In language modeling based music generation, a generated waveform is represented by a sequence of hierarchical token stacks that can be decoded either in an auto-regressive manner or in parallel, depending on the codebook patterns. In particular, flattening the codebooks represents the highest quality decoding strategy, while being notoriously slow. To this end, we propose a novel stack-and-delay style of decoding strategy to improve upon the flat pattern decoding where generation speed is four times faster as opposed to vanilla flat decoding. This brings the inference time close to that of the delay decoding strategy, and allows for faster inference on GPU for small batch sizes. For the same inference efficiency budget as the delay pattern, we show that the proposed approach performs better in objective evaluations, almost closing the gap with the flat pattern in terms of quality. The results are corroborated by subjective evaluations which show that samples generated by the new model are slightly more often preferred to samples generated by the competing model given the same text prompts.
In this work, we provide a comprehensive survey of AI music generation tools, including both research projects and commercialized applications. To conduct our analysis, we classified music generation approaches into three categories: parameter-based, text-based, and visual-based classes. Our survey highlights the diverse possibilities and functional features of these tools, which cater to a wide range of users, from regular listeners to professional musicians. We observed that each tool has its own set of advantages and limitations. As a result, we have compiled a comprehensive list of these factors that should be considered during the tool selection process. Moreover, our survey offers critical insights into the underlying mechanisms and challenges of AI music generation.
Music generation has attracted growing interest with the advancement of deep generative models. However, generating music conditioned on textual descriptions, known as text-to-music, remains challenging due to the complexity of musical structures and high sampling rate requirements. Despite the task's significance, prevailing generative models exhibit limitations in music quality, computational efficiency, and generalization. This paper introduces JEN-1, a universal high-fidelity model for text-to-music generation. JEN-1 is a diffusion model incorporating both autoregressive and non-autoregressive training. Through in-context learning, JEN-1 performs various generation tasks including text-guided music generation, music inpainting, and continuation. Evaluations demonstrate JEN-1's superior performance over state-of-the-art methods in text-music alignment and music quality while maintaining computational efficiency. Our demos are available at http://futureverse.com/research/jen/demos/jen1
In this work, we investigate the personalization of text-to-music diffusion models in a few-shot setting. Motivated by recent advances in the computer vision domain, we are the first to explore the combination of pre-trained text-to-audio diffusers with two established personalization methods. We experiment with the effect of audio-specific data augmentation on the overall system performance and assess different training strategies. For evaluation, we construct a novel dataset with prompts and music clips. We consider both embedding-based and music-specific metrics for quantitative evaluation, as well as a user study for qualitative evaluation. Our analysis shows that similarity metrics are in accordance with user preferences and that current personalization approaches tend to learn rhythmic music constructs more easily than melody. The code, dataset, and example material of this study are open to the research community.
Text-to-music generation (T2M-Gen) faces a major obstacle due to the scarcity of large-scale publicly available music datasets with natural language captions. To address this, we propose the Music Understanding LLaMA (MU-LLaMA), capable of answering music-related questions and generating captions for music files. Our model utilizes audio representations from a pretrained MERT model to extract music features. However, obtaining a suitable dataset for training the MU-LLaMA model remains challenging, as existing publicly accessible audio question answering datasets lack the necessary depth for open-ended music question answering. To fill this gap, we present a methodology for generating question-answer pairs from existing audio captioning datasets and introduce the MusicQA Dataset designed for answering open-ended music-related questions. The experiments demonstrate that the proposed MU-LLaMA model, trained on our designed MusicQA dataset, achieves outstanding performance in both music question answering and music caption generation across various metrics, outperforming current state-of-the-art (SOTA) models in both fields and offering a promising advancement in the T2M-Gen research field.
Autoregressive models based on Transformers have become the prevailing approach for generating music compositions that exhibit comprehensive musical structure. These models are typically trained by minimizing the negative log-likelihood (NLL) of the observed sequence in an autoregressive manner. However, when generating long sequences, the quality of samples from these models tends to significantly deteriorate due to exposure bias. To address this issue, we leverage classifiers trained to differentiate between real and sampled sequences to identify these failures. This observation motivates our exploration of adversarial losses as a complement to the NLL objective. We employ a pre-trained Span-BERT model as the discriminator in the Generative Adversarial Network (GAN) framework, which enhances training stability in our experiments. To optimize discrete sequences within the GAN framework, we utilize the Gumbel-Softmax trick to obtain a differentiable approximation of the sampling process. Additionally, we partition the sequences into smaller chunks to ensure that memory constraints are met. Through human evaluations and the introduction of a novel discriminative metric, we demonstrate that our approach outperforms a baseline model trained solely on likelihood maximization.
This paper presents an innovative approach to enhance control over audio generation by emphasizing the alignment between audio and text representations during model training. In the context of language model-based audio generation, the model leverages input from both textual and audio token representations to predict subsequent audio tokens. However, the current configuration lacks explicit regularization to ensure the alignment between the chosen text representation and the language model's predictions. Our proposal involves the incorporation of audio and text representation regularization, particularly during the classifier-free guidance (CFG) phase, where the text condition is excluded from cross attention during language model training. The aim of this proposed representation regularization is to minimize discrepancies in audio and text similarity compared to other samples within the same training batch. Experimental results on both music and audio generation tasks demonstrate that our proposed methods lead to improvements in objective metrics for both audio and music generation, as well as an enhancement in the human perception for audio generation.
As generative models have risen in popularity, a domain that has risen alongside is generative models for music. Our study aims to compare the performance of a simple Markov chain model and a recurrent neural network (RNN) model, two popular models for sequence generating tasks, in jazz music improvisation. While music, especially jazz, remains subjective in telling whether a composition is "good" or "bad", we aim to quantify our results using metrics of groove pattern similarity and pitch class histogram entropy. We trained both models using transcriptions of jazz blues choruses from professional jazz players, and also fed musical jazz seeds to help give our model some context in beginning the generation. Our results show that the RNN outperforms the Markov model on both of our metrics, indicating better rhythmic consistency and tonal stability in the generated music. Through the use of music21 library, we tokenized our jazz dataset into pitches and durations that our model could interpret and train on. Our findings contribute to the growing field of AI-generated music, highlighting the important use of metrics to assess generation quality. Future work includes expanding the dataset of MIDI files to a larger scale, conducting human surveys for subjective evaluations, and incorporating additional metrics to address the challenge of subjectivity in music evaluation. Our study provides valuable insight into the use of recurrent neural networks for sequential based tasks like generating music.
This paper introduces the HumTrans dataset, which is publicly available and primarily designed for humming melody transcription. The dataset can also serve as a foundation for downstream tasks such as humming melody based music generation. It consists of 500 musical compositions of different genres and languages, with each composition divided into multiple segments. In total, the dataset comprises 1000 music segments. To collect this humming dataset, we employed 10 college students, all of whom are either music majors or proficient in playing at least one musical instrument. Each of them hummed every segment twice using the web recording interface provided by our designed website. The humming recordings were sampled at a frequency of 44,100 Hz. During the humming session, the main interface provides a musical score for students to reference, with the melody audio playing simultaneously to aid in capturing both melody and rhythm. The dataset encompasses approximately 56.22 hours of audio, making it the largest known humming dataset to date. The dataset will be released on Hugging Face, and we will provide a GitHub repository containing baseline results and evaluation codes.
Large Language Models (LLMs) have shown immense potential in multimodal applications, yet the convergence of textual and musical domains remains relatively unexplored. To address this gap, we present MusiLingo, a novel system for music caption generation and music-related query responses. MusiLingo employs a single projection layer to align music representations from the pre-trained frozen music audio model MERT with the frozen LLaMA language model, bridging the gap between music audio and textual contexts. We train it on an extensive music caption dataset and fine-tune it with instructional data. Due to the scarcity of high-quality music Q&A datasets, we created the MusicInstruct (MI) dataset from MusicCaps, tailored for open-ended music inquiries. Empirical evaluations demonstrate its competitive performance in generating music captions and composing music-related Q&A pairs. Our introduced dataset enables notable advancements beyond previous ones.