Codec2

CODEC2 is a ( patent ) free lossy audio codec, which is dedicated to understand human speech transmission at very low bit rates. The codec was created to carry voice signals over narrowband radio channels in amateur radio. The reference implementation is subject to the terms of version 2.1 of the GNU Lesser General Public License ( LGPL).

The open specification of the process provides digital communication over amateur radio frequencies without transmitting through the use of currently available proprietary codecs such as AMBE or MELP necessarily unspecified digital content, which is prohibited radio amateurs.

CODEC2 has been officially integrated in FreeSWITCH and there is a patch for integration with Asterisk available.

Features

CODEC2 offers modes with a fixed bit rate of 3,200, 2,400, 1,400 or 1,200 bit / s It processes and provides the PCM data with a sampling frequency of 8 kHz. The individual (parameter ) data package covers 10 to 20 ( 2.4 kbit / s) and 40 (1.4 kbit / s) milliseconds. The algorithmic latency quantifies the author of about 100 milliseconds. The voice quality moves slightly below the ordinary 2G mobile phones and can allegedly measure at comparatively much lower bit rate with that of AMBE.

The reference implementation is written in C and is not yet without floating-point arithmetic of, the method does not require to be so. The reference software package also includes a FDMDV software modem and a graphical user interface based on FLTK. The software is developed on Linux, and it is next to a Linux version also created using Cygwin Windows port available.

Main developer Rowe avoided in principle algorithms that are affected by valid patents by building his process fundamentally on for decades known techniques. However, to the presentation on the linux.conf.au in January 2012, no comprehensive patent search has been performed.

Technology

The method employs the tools of parametric audio coding using a model of the human voice. It uses, among others, a sinusoidal model as a basic method, which is based on developments by Robert J. McAulay and Thomas F. Quatieri (MIT Lincoln labs ) from the mid-1980s, going back and closely with the Multi- Band Excitation codecs is related. Parameters for the description of line spectrum pairs (a type of LPC coefficients ), ( fundamental) pitch, energy and voicing of the signal are determined and quantized from the input signal. A PCM signal is synthesized from it again On the receiver side. The Sinusoidal model based on regularities ( periodicity ) in the pattern of overtone and layered harmonic sinusoids over a determined basic frequency. The amplitudes of the harmonics are modeled with the Linear Predictive Coding ( LPC).

History

The prominent free software advocate and radio amateur Bruce Perens saw the need for a free voice codec for under 5 kbit / s He said in 2008 Jean -Marc Valin ( Speex, Opus ) on it, which introduced him to the lead developer David Grant Rowe, who has worked variously with valine together to Speex. Rowe himself is also a radio amateur ( callsign VK5DGR ) and has experience with the creation and the use of codecs and other signal processing algorithms for speech signals. He has, among others, in the 1990s obtained a doctoral degree in speech coding and helped build one of the first satellite telephony systems ( Mobilesat ).

He was convinced by the task and was on 21 August 2009 announced its decision to want to work at an appropriate codec. He built on the research and findings of his doctoral thesis. In August 2010 he released version 0.1 alpha.

Towards the end of 2011 Version 0.2 was released, which introduces a mode with 1400 bits / s and brings substantial improvements in quantization.

In January 2012, as part of linux.conf.au Jean -Marc Valin helped with the improvement of quantization of the line spectrum pairs, bringing Rowe knows less. After several changes to the available bit rate modes in the winter and spring of 2011/2012 are since May modes with 2,400, 1,400 and 1,200 bit / s available.