Comfort noise is the quiet, synthetic background hiss a receiver generates to fill the silence when discontinuous transmission has stopped sending real packets. Without it, the gaps during DTX would be dead digital silence, and listeners reliably interpret total silence as a dropped call, so they start saying 'hello? are you there?'. The encoder sends a tiny silence descriptor (a SID frame, standardized as RFC 3389 in RTP) describing the spectral shape and level of the real background noise, and the decoder synthesizes a matching low-level noise so the transition in and out of silence is seamless. It is a small touch that makes bandwidth-saving DTX feel natural rather than broken.
The book · Volume 4 of 7
Audio for Video: A Complete Guide to Sound in Video Products: From PCM and Codecs to Loudness, Accessibility, and Real-Time Voice Processing
Comfort noise keeps silence from sounding like a dropped call, and mismatched levels make it a source of complaints of its own. Volume 4 covers comfort noise in detail within real-time voice processing: VAD and discontinuous transmission, speech codecs and Opus, noise suppression and how these stages interact on a call. Written by Nikolay Sapunov, CEO at Fora Soft.
Users thinking calls dropped during silence despite comfort noise?
Send us call recordings and codec settings, and we will tune comfort noise, VAD and DTX so silence sounds natural. 250+ video products shipped since 2005.
