论文解读
为说话人日志做自监督：目标、数据与容量的分工与代价
该文以 WavLM 为骨干研究说话人日志，报告加权多层目标、10k 小时数据规模、Large 容量与 Conformer 切块调优带来可复现的 DER 下降，但 100k 小时与剪枝收益趋平且标注噪声限制可测改进。
预训练
论文解读
说话人受控下做语种识别：用任务相关预训练压住说话人捷径，再用间隔损失拉开语种边界
针对同一说话人说多种语言时模型易依赖说话人特征走捷径的问题，该文采用在 VoxLingua107 上预训练的 ECAPA-TDNN 做编码器并比较 AAM-Softmax 与 RAM-Softmax，在 Tidy-X 上报告识别任务 85.95% 宏准确率与验证任务 16.39% 等错误率，最强证据来自与官方基线的同协议对照，代价是未充分优化未见语种验证与未 …
论文解读
不用真谱也能学会听谱：TUTTI 用 36 万首合成曲喂饱纯 Transformer
音乐转录 | 7.4/10
论文解读
32 分钟语料能给 Baniwa ASR 一个起点，却不能借此跳过泛化检验
语音识别 | 4.7/10
论文解读
LAION-BVD: A 10-Million-Hour Open Video Dataset for Multimodal Pre-training
音视频理解 | 9.4/10
论文解读
Do Time-Series Foundation Models Pay Off for Industrial Monitoring? A Cost-Aware Empirical Study
音频事件检测 | 8.8/10
论文解读
Listening Forward: Next Patch Embedding Prediction Enables Scalable Audio Learners
音频理解 | 8.8/10
论文解读
Alignment Is All You Need: Instruction-Free Training for General Audio-Language Models
音频理解 | 8.0/10
论文解读
Target Speaker Identification: A Low-Latency Streaming Pipeline
说话人验证 | 5.6/10
论文解读
An Analytical-Prior Framework for Data-Efficient Prediction of Sound-Reduction Frequencies in Rectangular Side-Branch Helmholtz Resonators
预训练 | 5.8/10
论文解读
Navigating Speech Enhancement for Real-Time MRI: A Systematic Assessment of Signal Quality, Source Preservation, and Downstream Tasks
语音增强 | 5.7/10
论文解读
Evaluating Pre-trained Speech Encoders for Spontaneous Speech Detection and Out of Domain Synthetic Speech Generalisation in Indic Languages
语音伪造检测 | 6.0/10
论文解读
Luna-TTS Family Technical Report
语音合成 | 6.9/10
论文解读
DINO-A: Adapting Self-Distillation Vision Transformers to General Audio Representation Learning
音频分类 | 6.2/10
论文解读
Audio-to-Score Transcription using Pre-trained Features, Data Augmentation, and the New SheetSage-A2S Dataset
音乐转录 | 7.7/10
论文解读
Assessing speech quality metrics for evaluation of neural audio codecs under clean speech conditions
语音质量评估 | 6.0/10
论文解读
MERaLiON-GR: Speech Gender Recognition Model for English and SEA Languages
语音属性识别 | 7.5/10
论文解读
PASE: Leveraging the Phonological Prior of WavLM for Low-Hallucination Generative Speech Enhancement
语音增强 | 8.1/10
论文解读
Can Foundation Models Hear What Made That Sound? A Tiered Benchmark of Audio-Language Models and Traditional Classifiers for Closed-Set Sound Source Identification
音频分类 | 6.3/10
论文解读
Self-Supervised Audio Representation Learning for Pediatric Asthma Detection in Emergency Care Using Digital Stethoscope Recordings
音频分类 | 5.0/10