Audio Codecs
elemedia® Speech and Audio Codecs
Agora Labs offers software components in support of application developers in the multimedia over IP industry. Agora Labs' customers typically build upon these components to develop new multimedia services or applications.
Transmission of audio in its raw format may be bandwidth intensive. The existing telephone network only transmits about 3.4 KHz worth of speech data, and compands it by means of either u-Law (in North America and Japan) or A-Law (in much of the rest of the world) so it can travel over a 64 Kbit/sec channel. Some codecs allow a wider range of input frequency (e.g. G.722 allows 7 KHz), and most codecs compress the data to less than 64 Kbits/sec.
Like protocol stacks, video codecs that adhere to the ITU-T international standards insure complete interoperability with other products. Agora Labs has two ITU-T video codecs available, a G.711 codec and a G.722 codec.
G.711 specifies Pulse code modulation (PCM) of voice frequencies; it uses the same companding scheme the existing telephone network uses. G.711 is a mandatory codec for H.323 systems. G.711 has low processor utilization, but has a relatively high IP bandwidth requirement.
G.722 specifies 7 kHz analog audio-coding within 64 kbit/s, and is often used in videoconferencing systems where a higher audio quality is required. Most of the energy in speech is within this range.
Low and Ultra Low Bitrate Codecs -- Agora has added support for two of elemedia's low-bitrate codecs. While these are not toll quality, they sound surprisingly good.
G.723.1 and G.729A -- Agora has added support for G.732.1 and G.729A.
The SX7300P provides a high performance, low cost, low complexity solution ideal for speech transmission over the Internet or other packet networks. The coder offers close to toll quality speech and, with a rate of just 7.3 kbit/s, enables speech transmission over either V.34 or V.32bis modems. At V.34 data rates multiple speech channels or multimedia applications such as simultaneous voice and data may be supported on a single telephone line. The small frame size of the coder provides low end-to-end delay. A powerful frame reconstruction algorithm yields exceptionally robust performance in the presence of channel errors. The low delay characteristics and good tandeming performance make it ideal for multiparty conferencing. Modest memory and processing power requirements enable implementations of the coder in either Pentium or 486 environments or on inexpensive DSPs. The SX7300P offers speech quality comparable with ITU G.723 and G.729 Annex A with enhanced performance in the presence of lost speech frames and at significantly lower complexity and lower cost. This codec has its origins in Lucent's Bell Labs.
G.729 and G.729A patent licensing: these codecs require licensing from NTT, University of Sherbrooke, France Telecom, Nokia, NEC, Mitsubushi and from Lucent.
G.723.1 patent licensing: G.723.1 requires a license from Audiocodes Ltd, France Telecom, University of Sherbrooke, Cisco, NTT, DSP Group, Nokia and Lucent.
The SX7300P only requires a license from Lucent, which has the same price as the Lucent G.723.1 or G.729A patent licenses (and the SX7300P patent royalty is less costly than the Lucent G.729 patent royalty.) By using the SX7300P, you will save a lot on patent licensing.
The SX9600P provides true toll quality at a bit rate of 9.6 kbit/s making it ideal for a broad range of applications. For corporate intranet or extranet applications where quality needs to be preserved, the SX9600P provides quality that is superior to the SX7300P. For general internet applications, the coder provides excellent quality while still enabling speech transmission over either V.34 or V.32bis modems. At V.34 data rates, multiple speech channels or multimedia applications such as simultaneous voice and data may be supported on a single telephone line. The small frame size of the coder provides low end-to-end delay. A powerful frame reconstruction algorithm yields exceptionally robust performance in the presence of channel errors. The low delay characteristics and good tandeming performance make it ideal for multiparty conferencing. Modest memory and processing power requirements enable implementations of the coder in either PentiumTM or 486TM environments or on inexpensive DSPs. This codec has its origins at Lucent's Bell Labs, and only requires an IP royalty payment to Lucent (this royalty is the same amount as the SX7300P IP royalty.)
Customers normally license these products as object code software, they are available on the Pentium processor (IBM PC), PowerPC, MIPS, ARM and several popular DSP families.
Agora Labs is adding to their speech and audio codec product line, contact us if you don't see the codec you need!