Over a year and a half after the previous major 1.5 release, Opus, a free, open-source audio codec designed for real-time communication and high-quality audio streaming, widely used for voice, music, and interactive applications, has released version 1.6.
A major update is a new wideband-to-fullband bandwidth extension module. This machine learning component helps rebuild higher-frequency sounds, making audio clearer when the source or network quality is limited. It builds on the ML work from the last release and aims for a more natural, consistent sound.
Opus 1.6 also introduces support for 96 kHz audio in Opus HD, extending the codec’s high-resolution capabilities. This makes Opus more suitable for professional and high-fidelity use cases that require higher sampling rates, while maintaining its efficiency and flexibility across a wide range of bitrates.
Moreover, the new version is now more reliable on unstable networks. Deep Redundancy, which helps keep audio smooth during packet loss, has been improved, reducing sound glitches and dropouts, especially during real-time streaming and calls.
For developers, Opus 1.6 adds a new 24-bit encoder and decoder API, enabling higher-precision audio processing where needed. The release also includes several fixed-point optimizations, which are especially relevant for embedded systems and platforms without efficient floating-point hardware.
Beyond these headline features, Opus 1.6 incorporates numerous minor bug fixes and stability improvements. For more information, see the announcement.
