arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2606.01009cs.SD
收藏

MelT: 面向现代加速器的高效单级音频前端的GEMM原生NDFT

MelT: A Portable, Single-GEMM Mel Audio Frontend via Non-Uniform DFT with Measured Latency and Energy Gains on GPUs

Instituto de Ciências Matemáticas e de Computação, University of São Paulo, Brazil(圣保罗大学数学与计算机科学研究所,巴西)

Augusto Camargo

AI总结:

提出MelT框架,通过将梅尔间隔非均匀离散傅里叶变换(NDFT)公式化为稠密通用矩阵乘法(GEMM)操作,实现单级音频前端,替代传统STFT+梅尔流水线,在多种加速器上获得高达3.75倍推理加速和3.52倍能耗降低。

摘要:

现代音频处理网络通常部署在加速器上,其峰值吞吐量通过稠密线性代数获得,而传统的声学前端——短时傅里叶变换(STFT)后接稀疏梅尔聚合——在结构上仍然是异构的。这种不匹配会在当代加速器后端引入内存带宽、调度和中间分配开销。本文介绍MelT,一个单级前端框架,其中梅尔间隔非均匀离散傅里叶变换(NDFT)基被预先计算,并通过稠密通用矩阵乘法(GEMM)操作应用于时域声学帧。贡献不在于NDFT算子本身,而在于将梅尔间隔NDFT投影公式化为GEMM原生的音频前端,并将其评估为传统STFT+梅尔流水线的硬件高效替代方案。在从Apple A18 Pro边缘硬件到NVIDIA H100数据中心加速器的多个平台上评估,MelT在保持下游分类准确性的同时,实现了高达3.75倍的推理延迟加速和3.52倍的能耗降低。

英文摘要:

Modern neural audio models run on accelerators whose peak throughput comes from dense matrix multiplication, increasingly at the edge and in datacenters. The conventional acoustic frontend, however -- a Short-Time Fourier Transform (STFT) followed by sparse Mel aggregation -- remains a multi-stage pipeline centered on the Fast Fourier Transform (FFT), with execution overheads unlike the dense linear algebra dominating the inference stack. This work introduces MelT, a portable single-stage Mel frontend that precomputes Mel-spaced Non-Uniform Discrete Fourier Transform (NDFT) bases and applies them to time-domain frames through General Matrix Multiplication (GEMM). The contribution is a computational design principle: decoupling Mel feature extraction from vendor-specific FFT primitives and lowering it onto the matrix-multiplication substrate accelerators already optimize. It is not a new spectral operator. MelT's direct projection performs more arithmetic than the FFT pipeline. Yet in the compact-resolution regime of neural audio frontends, it achieves a 1.64-times to 3.29-times latency reduction and up to a 3.03-times reduction in measured active energy, from the Apple A18 Pro to the NVIDIA H100. All gains are within-platform comparisons, accompanied by task-level validation. Word error rate stays statistically equivalent to the native frontend's on frozen Whisper models of medium size and larger; speaker-attribute classification on VoxCeleb1 is non-inferior. The cepstral extension MFCCT preserves utility on a clinical respiratory-insufficiency classification task (SPIRA) while improving on the MFCC baseline. These results indicate that, in practical regimes on the accelerators evaluated here, hardware alignment rather than arithmetic count can govern the realized cost of feature extraction.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑