Investigating Design Choices in Joint-Embedding Predictive Architectures for General Audio Representation Learning

• URL: http://arxiv.org/abs/2405.08679v1
• Date: Tue, 14 May 2024 15:00:09 GMT
• Title: Investigating Design Choices in Joint-Embedding Predictive Architectures for General Audio Representation Learning
• Authors: Alain Riou, Stefan Lattner, Gaëtan Hadjeres, Geoffroy Peeters,
• Abstract summary: This paper addresses the problem of self-supervised general-purpose audio representation learning. We explore the use of Joint-Embedding Predictive Architectures (JEPA) for this task, which consists of splitting an input mel-spectrogram into two parts (context and target), computing neural representations for each, and training the neural network to predict the target representations from the context representations.
• Score: 3.7161123856095837
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: This paper addresses the problem of self-supervised general-purpose audio representation learning. We explore the use of Joint-Embedding Predictive Architectures (JEPA) for this task, which consists of splitting an input mel-spectrogram into two parts (context and target), computing neural representations for each, and training the neural network to predict the target representations from the context representations. We investigate several design choices within this framework and study their influence through extensive experiments by evaluating our models on various audio classification benchmarks, including environmental sounds, speech and music downstream tasks. We focus notably on which part of the input data is used as context or target and show experimentally that it significantly impacts the model's quality. In particular, we notice that some effective design choices in the image domain lead to poor performance on audio, thus highlighting major differences between these two modalities.

Related papers

• AV-SUPERB: A Multi-Task Evaluation Benchmark for Audio-Visual Representation Models [92.92233932921741]
  We propose the AV-SUPERB benchmark that enables general-purpose evaluation of unimodal audio/visual and bimodal fusion representations. We evaluate 5 recent self-supervised models and show that none of these models generalize to all tasks. We show that representations may be improved with intermediate-task fine-tuning and audio event classification with AudioSet serves as a strong intermediate task.
  arXiv  Detail & Related papers  (2023-09-19T17:35:16Z)
• Improving Audio-Visual Segmentation with Bidirectional Generation [40.78395709407226]
  We introduce a bidirectional generation framework for audio-visual segmentation. This framework establishes robust correlations between an object's visual characteristics and its associated sound. We also introduce an implicit volumetric motion estimation module to handle temporal dynamics.
  arXiv  Detail & Related papers  (2023-08-16T11:20:23Z)
• Unraveling Instance Associations: A Closer Look for Audio-Visual Segmentation [18.001730255429347]
  Audio-visual segmentation (AVS) is a challenging task that involves accurately segmenting sounding objects based on audio-visual cues. We propose a new cost-effective strategy to build challenging and relatively unbiased high-quality audio-visual segmentation benchmarks. Experiments conducted on existing AVS datasets and on our new benchmark show that our method achieves state-of-the-art (SOTA) segmentation accuracy.
  arXiv  Detail & Related papers  (2023-04-06T09:54:06Z)
• An Empirical Investigation of Commonsense Self-Supervision with Knowledge Graphs [67.23285413610243]
  Self-supervision based on the information extracted from large knowledge graphs has been shown to improve the generalization of language models. We study the effect of knowledge sampling strategies and sizes that can be used to generate synthetic data for adapting language models.
  arXiv  Detail & Related papers  (2022-05-21T19:49:04Z)
• Deep Learning-based Non-Intrusive Multi-Objective Speech Assessment Model with Cross-Domain Features [30.57631206882462]
  The MOSA-Net is designed to estimate speech quality, intelligibility, and distortion assessment scores based on a test speech signal as input. We show that the MOSA-Net can precisely predict perceptual evaluation of speech quality (PESQ), short-time objective intelligibility (STOI), and speech distortion index (BLS) scores when tested on both noisy and enhanced speech utterances.
  arXiv  Detail & Related papers  (2021-11-03T17:30:43Z)
• LDNet: Unified Listener Dependent Modeling in MOS Prediction for Synthetic Speech [67.88748572167309]
  We present LDNet, a unified framework for mean opinion score (MOS) prediction. We propose two inference methods that provide more stable results and efficient computation.
  arXiv  Detail & Related papers  (2021-10-18T08:52:31Z)
• Data Fusion for Audiovisual Speaker Localization: Extending Dynamic Stream Weights to the Spatial Domain [103.3388198420822]
  Esting the positions of multiple speakers can be helpful for tasks like automatic speech recognition or speaker diarization. This paper proposes a novel audiovisual data fusion framework for speaker localization by assigning individual dynamic stream weights to specific regions. A performance evaluation using audiovisual recordings yields promising results, with the proposed fusion approach outperforming all baseline models.
  arXiv  Detail & Related papers  (2021-02-23T09:59:31Z)
• Audiovisual Highlight Detection in Videos [78.26206014711552]
  We present results from two experiments: efficacy study of single features on the task, and an ablation study where we leave one feature out at a time. For the video summarization task, our results indicate that the visual features carry most information, and including audiovisual features improves over visual-only information. Results indicate that we can transfer knowledge from the video summarization task to a model trained specifically for the task of highlight detection.
  arXiv  Detail & Related papers  (2021-02-11T02:24:00Z)
• An Empirical Study of Visual Features for DNN based Audio-Visual Speech Enhancement in Multi-talker Environments [5.28539620288341]
  AVSE methods use both audio and visual features for the task of speech enhancement. To the best of our knowledge, there is no published study that has investigated which visual features are best suited for this specific task. Our study shows that despite the overall better performance of embedding-based features, their computationally intensive pre-processing make their use difficult in low resource systems.
  arXiv  Detail & Related papers  (2020-11-09T11:48:14Z)
• COALA: Co-Aligned Autoencoders for Learning Semantically Enriched Audio Representations [32.456824945999465]
  We propose a method for learning audio representations, aligning the learned latent representations of audio and associated tags. We evaluate the quality of our embedding model, measuring its performance as a feature extractor on three different tasks.
  arXiv  Detail & Related papers  (2020-06-15T13:17:18Z)
• Dynamic Feature Integration for Simultaneous Detection of Salient Object, Edge and Skeleton [108.01007935498104]
  In this paper, we solve three low-level pixel-wise vision problems, including salient object segmentation, edge detection, and skeleton extraction. We first show some similarities shared by these tasks and then demonstrate how they can be leveraged for developing a unified framework.
  arXiv  Detail & Related papers  (2020-04-18T11:10:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.