Research map / Architectures and efficiency / Attention and transformers
Multi-granularity Interactive Attention Framework for Residual Hierarchical Pronunciation Assessment
Hong Han, Hao-Chen Pei, Zhao-Zheng Nie, Xin Luo, Xin-Shun Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add interactive attention heads at all granularities (ablation) | HIA · speechocean762 | word stress PCC: 0.335 -> 0.436 | Improved |
| add interactive attention heads at all granularities (ablation) | HIA · speechocean762 | utterance total PCC: 0.748 -> 0.764 | Improved |
| add residual connection from encoder embeddings (ablation) | HIA · speechocean762 | word stress PCC: 0.382 -> 0.436 | Improved |
| add hierarchical score passing between granularities (ablation) | HIA · speechocean762 | word stress PCC: 0.374 -> 0.436 | Improved |
| add one convolutional layer per granularity (ablation) | HIA · speechocean762 | word total PCC: 0.601 -> 0.628 | Improved |
| increase convolutional layers from 1 to 2 or 3 (ablation) | HIA · speechocean762 | utterance total PCC: 0.764 -> 0.759 (2 layers), 0.755 (3 layers) | Hurt |
| increase embedding size from 48 to 96 (ablation) | HIA · speechocean762 | word total PCC: 0.628 -> 0.611 | Hurt |
| reduce embedding size from 48 to 24 (ablation) | HIA · speechocean762 | utterance total PCC: 0.764 -> 0.752 | Hurt |
| use multiple attention heads instead of one (ablation) | HIA · speechocean762 | utterance total PCC: 0.764 -> 0.759 (2 heads), 0.751 (4 heads) | Hurt |
| HIA vs prior multi-granularity SOTA Gradformer | HIA · speechocean762 | phoneme MSE: 0.079 -> 0.076 | Improved |
| HIA vs single-granularity LAS on utterance total | HIA · speechocean762 | utterance total PCC: 0.766 (LAS) vs 0.764 | Hurt |
Closest papers by meaning
- EmergentTTS-Eval: Evaluating TTS Models on Complex Prosodic, Expressiveness, and Linguistic Challenges Using Model-as-a-Judge (NeurIPS 2025 · similarity 0.83)
- MMSU: A Massive Multi-task Spoken Language Understanding and Reasoning Benchmark (ICLR 2026 · similarity 0.83)
- Objective Soups: Multilingual Multi-Task Modeling for Speech Processing (NeurIPS 2025 · similarity 0.83)
- Towards Multi-Intent Spoken Language Understanding via Hierarchical Attention and Optimal Transport (AAAI 2024 · similarity 0.82)
- OWLS: Scaling Laws for Multilingual Speech Recognition and Translation Models (ICML 2025 · similarity 0.82)
- Toward Adaptive Large Language Models Structured Pruning via Hybrid-grained Weight Importance Assessment (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
