Noise Suppression Evaluation of Virtual Communication Platforms

Effective background noise suppression is critical for ensuring clear communication during video and audio calls. We evaluated the noise suppression capabilities of major virtual communication platforms by conducting a comprehensive study using ITU-T P.835 standard. The analysis combined subjective listening tests with objective DNSMOS-based evaluations to assess real-world noise scenarios.

Noise suppression effectiveness evaluated by these categories:

• MOS_BAK - Intrusiveness of the background noise
• MOS_SIG - Distortion of the speech signal
• MOS_OVRL - Overall quality of the speech sample

Key takeaways:

• MS Teams achieved the highest overall quality, balancing strong noise reduction with preservation of speech clarity.
• Several competitors showed similar background noise suppression (MOS_BAK) levels in objective DNSMOS evaluations, indicating effective noise reduction capabilities across platforms.
• Differences between subjective and objective evaluations suggest DNSMOS ratings were more conservative compared to listener impressions.
• All tested platforms showed significant improvements compared to the noisy baseline, with some competitors performing on par with MS Teams for background noise handling.