Research map / Label-efficient and robust learning / Noisy labels
GOMPSNR: Reflourish the Signal-to-Noise Ratio Metric for Audio Generation Tasks
Lingling Dai, Andong Li, Cheng Chi, Yifan Liang, Xiaodong Li, Chengshi Zheng
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 |
|---|---|---|---|
| replace IP with omnidirectional phase derivatives in SNR (OMPSNR) (ablation) | SNR · LibriTTS, pretrained Vocos | PCC/SRCC with perceptual metrics: SNR PCC/SRCC not exceeding 0.1 -> significantly higher | Improved |
| rectify correlation component (GOMPSNR vs OMPSNR) (ablation) | OMPSNR · LibriTTS, pretrained Vocos | PCC/SRCC with perceptual metrics | Improved |
| add vanilla phase loss P (ablation) | Vocos · LJSpeech | PESQ: 3.749 -> 3.711 | No clear effect |
| add OP loss (ablation) | Vocos · LJSpeech | PESQ: 3.749 -> 3.752 | No clear effect |
| weight OP loss by magnitude (WOP) (ablation) | Vocos · LJSpeech | PESQ: 3.752 -> 3.928 | Improved |
| add vanilla RI loss on top of WOP (ablation) | Vocos + WOP · LJSpeech | PESQ: 3.928 -> 3.891 | Hurt |
| replace RI with OmniRI (L1) (ablation) | Vocos + WOP · LJSpeech | PESQ: 3.891 -> 3.998 | Improved |
| use CORI instead of ORI (L1) (ablation) | Vocos + WOP · LJSpeech | PESQ: 3.998 -> 3.992 | No clear effect |
| use L2 instead of L1 distance in ORI (ablation) | Vocos + WOP + ORI · LJSpeech | PESQ: 3.998 -> 3.791 | Hurt |
| linear magnitude loss instead of log (ablation) | APNet · LJSpeech | PESQ / GOMPSNR: PESQ 2.836 -> 3.124; GOMPSNR 3.335 -> -0.957 | Mixed |
| linear magnitude loss instead of log (ablation) | Vocos · LibriTTS | PESQ: 3.167 -> 3.298 | Improved |
| full proposed loss combination (Lin, WOP, CORI) | APNet2 · LibriTTS | PESQ: 1.685 -> 3.789 | Improved |
Closest papers by meaning
- Audio Super-Resolution with Latent Bridge Models (NeurIPS 2025 · similarity 0.84)
- Vocos: Closing the gap between time-domain and Fourier-based neural vocoders for high-quality audio synthesis (ICLR 2024 · similarity 0.83)
- AGAV-Rater: Adapting Large Multimodal Model for AI-Generated Audio-Visual Quality Assessment (ICML 2025 · similarity 0.82)
- FlowDec: A flow-based full-band general audio codec with high perceptual quality (ICLR 2025 · similarity 0.82)
- V2A-Mapper: A Lightweight Solution for Vision-to-Audio Generation by Connecting Foundation Models (AAAI 2024 · similarity 0.82)
- Massive Sound Embedding Benchmark (MSEB) (NeurIPS 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.
