Effects of automotive microphone frequency response characteristics on speech and ASR quality before and after noise reduction: a continuous evaluation

Buccoli M, Du Y, Soendergaard J, Cazzaniga SS. Effects of automotive microphone frequency response characteristics on speech and ASR quality before and after noise reduction: a continuous evaluation. In: 2026 AES 6th International Automotive Audio Conference. Audio Engineering Society; 2026. Paper 17. Available from: https://aes.org/publications/elibrary-page/?id=23410

@inproceedings{Buccoli2026_23410,
  author = {Buccoli, Michele and Du, Yu and Soendergaard, Jacob and Cazzaniga, Simone Shawn},
  title = {{Effects of automotive microphone frequency response characteristics on speech and ASR quality before and after noise reduction: a continuous evaluation}},
  booktitle = {2026 AES 6th International Automotive Audio Conference},
  note = {Paper 17},
  year = {2026},
  month = jul,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=23410}
}

TY  - CPAPER
TI  - Effects of automotive microphone frequency response characteristics on speech and ASR quality before and after noise reduction: a continuous evaluation
AU  - Buccoli, Michele
AU  - Du, Yu
AU  - Soendergaard, Jacob
AU  - Cazzaniga, Simone Shawn
T2  - 2026 AES 6th International Automotive Audio Conference
M1  - Paper 17
PY  - 2026
DA  - 2026/07/29
UR  - https://aes.org/publications/elibrary-page/?id=23410
PB  - Audio Engineering Society
LA  - en
AB  - Microphones used in automotive hands-free and Automatic Speech Recognition (ASR) systems are typically required to meet wideband or fullband specifications defined by standards such as ITU-T P.1110 and P.1120. In practice, however, compliance with these standards is often challenged by vehicle cabin integration constraints and automotive-grade durability requirements. Moreover, there is limited empirical evidence clarifying how specific microphone characteristics influence perceptual audio quality and ASR performance. This paper presents an experimental investigation into the effects of microphone frequency response and bandwidth variations on system-level performance in automotive environments. Noise signals recorded under real-world driving conditions are used to evaluate perceptual speech quality using ETSI TS 103 281 metrics, including S-MOS, N-MOS, and G-MOS, as well as ASR accuracy quantified by Word Error Rate (WER). In addition, the interaction effects between microphone characteristics and signal processing algorithms are also studied by processing the signals through noise reduction (NR) modules. The results aim to identify the most critical frequency response attributes for properly determining microphone specifications in automotive hands-free and ASR applications.
ER  -