Arjun Bahuguna https://arjbah.github.io/ Recent content on Arjun Bahuguna Arjun Bahuguna https://arjbah.github.io/images/papermod-cover.png https://arjbah.github.io/images/papermod-cover.png Hugo en 2026 Arjun Bahuguna Wed, 27 May 2026 00:00:00 +0000 Embedding Geometry in Pretrained CLAP Models https://arjbah.github.io/posts/2026_06_evaluating_jtae/ Wed, 27 May 2026 00:00:00 +0000 https://arjbah.github.io/posts/2026_06_evaluating_jtae/ A long-form walkthrough of why pretrained CLAP models fail on compositional audio–text benchmarks, framed through the geometry of their embedding spaces: RankMe effective rank, Two-NN intrinsic dimension, residual singular-value energy, and the audio–text modality gap. Top 5 Takeaways from ICASSP 2026 https://arjbah.github.io/posts/2026_05_icassp_top10/ Sat, 09 May 2026 00:00:00 +0000 https://arjbah.github.io/posts/2026_05_icassp_top10/ Top 5 Takeaways from pre-ICASSP Workshop 2026 at MTG https://arjbah.github.io/posts/2026_05_preicassp_at_mtg/ Sun, 03 May 2026 00:00:00 +0000 https://arjbah.github.io/posts/2026_05_preicassp_at_mtg/ Evaluation of Music Generation Systems https://arjbah.github.io/posts/2026_01_evaluation_of_music_generation_system/ Thu, 01 Jan 2026 00:00:00 +0000 https://arjbah.github.io/posts/2026_01_evaluation_of_music_generation_system/ <p>The benchmarking of generative music systems represents a significant challenge in contemporary Music Information Retrieval because the field lacks a definitive ground truth against which synthetic outputs can be measured. Generative models such as those utilizing Transformer architectures or WaveNet variants often produce compositions that possess local coherence but fail to demonstrate global structural regularity or long-term repetitive dependencies (<a href="https://arjbah.github.io/posts/2026_01_evaluation_of_music_generation_system/#references">Wang et al., 2023</a>). Because artistic output is inherently subjective, the evaluation framework must transition beyond simple error minimization tasks to integrate multifaceted metrics that account for audio fidelity, musical theory adherence, and human perceptual experience (<a href="https://arjbah.github.io/posts/2026_01_evaluation_of_music_generation_system/#references">Lerch et al., 2025</a>).</p>