The irreversible voice compression codec "Opus" realizing ultra-low delay of 5 milliseconds for free, and what is amazing

The open source irreversible audio compression format "Opus" has performed a major update and the version has become "libopus 1.2". Based on changes and new functions from the previous version, what kind of voice compression codec is integral with Opus which features such as being able to realize low latency of 5 ms (milliseconds) for free, and so on.

Libopus 1.2 - Opus Codec
https://www.opus-codec.org/release/stable/2017/06/20/libopus-1_2.html


◆ What is Opus?
Opus is a versatile voice compression codec that can be used fully open & royalty free. Opus is suitable for Internet calling, music streaming, streaming application, etc., and Skype's "SILK Codec" and Xiph.Org Foundation's "CELT codecAs a standard "RFC 6716" incorporating the technology of "Internet Engineering Task Force(IETF) standardized by IETF.

◆ Function of Opus
OpusVoice over IPIt can handle a wide range of audio applications such as video conferencing, in-game audio chat, remote music live, etc. It can handle a wide range of audio applications from low bit rate (minimum 5 ms) narrow band audio to high quality stereo music To By using Opus with very low latency, it becomes possible to seamlessly change the streaming of voice and so on.

Opus supports the following functions.

· Bit rate: 6 KB / s to 510 KB / s
Sampling rate: 8 kHz (narrow band) to 48 kHz (full band)
Frame size: 2.5 ms to 60 ms
·Constant bit rate(CBR) ~Variable bit rate(VBR) support
· Audio frequency band: narrow band to full band
· Supports speech and music
· Supports stereo and mono
Supports up to 255 audio channels (multi-stream frame)
· Dynamically adjust bit rate, audio frequency band, frame size
· Less lossRobustness, Interpolation processing (PLC) with less packet loss
· Implementation of floating point and fixed point

◆ New features of libopus 1.2
· Speech quality improved especially at 12 to 20 kbit / s
Improved variable bit rate performance in hybrid mode
• Use of a wide range of speech frequency bands, including full-band speech starting at 14 kbit / s
· Music performance improved at 32 to 48 kb / s
· Generic ·SSE· CELT optimization
· Supports direct encoding of packets up to 120 ms
· In CELT modeDiscontinuous transfer(DTX) support
· Improved performance of SILK CBR
· Bug fixes

◆ Demo sample
On the following pages, samples are available to actually experience how much has changed. In the "Music Samples", and five of the codec (MP3 · Opus 1.0 · Opus 1.1 · Opus 1.2 · Uncompressed), switch the four types of bit rate (32kb / s · 48kb / s · 64kb / s · 96kb / s) It is possible to distinguish the difference. In "Speech. Samples", is switched five codecs (Speex · Opus 1.0 · Opus 1.1 · Opus 1.2 · Uncompressed), 4 types of bit rate (12kb / s · 16kb / s · 24kb / s · 32kb / s) It has become like.

Opus 1.2 Released
https://people.xiph.org/~jm/opus/opus-1.2/