SILK is the speech-optimized half of the Opus codec, contributed by Skype, where it originally powered Skype calls. It uses linear-predictive coding (LPC), the classic approach for voice, modelling the vocal tract to represent speech compactly and efficiently at low bitrates from narrowband to wideband. Opus automatically engages SILK (or a hybrid of SILK and CELT) when the content is speech and the bitrate is modest, which is most of the time on a voice call. Its presence is a big reason Opus sounds so good for conversation at rates where music codecs would struggle, while the CELT layer takes over for music and very low latency.
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
SILK is the half of Opus that carries speech at low bitrates, and knowing when Opus switches to it explains much of how calls sound on weak links. This volume covers SILK in detail inside Opus, with CELT, bandwidth modes, DTX and in-band FEC, so you can set voice bitrates with confidence. Written by Nikolay Sapunov, CEO at Fora Soft.
Tuning Opus SILK mode for clear voice on low-bandwidth connections?
Share your network targets and we will recommend Opus SILK and hybrid settings with listening tests at your target bitrates. 250+ video products shipped since 2005.
