User:Vinayr rao/Wideband Codec
Wideband Codecs in Digital telephony refers to the use of higher sampling rates than Narrowband Codecs or the utilization of embedded sub-band coding information to effectively increase the bandwidth of the baseband voice, from the traditional 200 Hz to 3.5 kHz used in Narrowband Codecs, to 50 Hz at the low end and anywhere from 7kHz to 22 kHz [1]. at the high end, depending on the type of codec used. This results in a significant improvement in voice quality.
Introduction:
Traditional analog telephone systems were designed to use a bandwidth of approximately 200 Hz to 3300 Hz as this was considered adequate for transmitting human voice intelligibly over a telephone system. When digital telephone systems replaced the earlier analog telephone systems, voice was sampled at a rate of 8 kHz with an 8-bit depth. According to the Nyquist-Shannon sampling theorem, choosing an 8 kHz sampling rate, would be sufficient to perfectly reproduce baseband information of up to 4 kHz. Hence an 8 kHz sampling rate was deemed sufficient to digitize voice for telephony.
The use of 8-bit depth would normally yield a theoretical Signal to Quantization Noise Ratio (SQNR of 49.93[2] dB for a pure sine-wave. However, with the use of companding (A-Law or u-law) it is possible to encode the 13-bit or 14-bit signed linear PCM samples respectively into logarithmic 8-bit samples. This achieves a better SQNR albeit at the expense of higher Total Harmonic Distortion (THD). Even so the voice quality of the G.711 PCM codec was deemed to be "toll quality" and was considered to be high quality in telephone systems.
Modern digital telecommunications systems are no longer limited by the constraints of the older systems. Wideband Codecs typically use one or more of the following methods to encode so called "Wideband voice".
Sampling Rate
Doubling the sampling rate, allows sampling of signals as high as 8 kHz. This allows the wideband codec to transmit consonants, sibilants and other subtleties of the human voice formerly lost or clipped by narrowband codecs and significantly adds to the Intelligibility and quality of the speech signal.
Bit-depth (Quantization)
The ubiquitous PCM codec, the venerable ITU-T G.711 samples at the rate of 8 kHz and after companding (A-Law or u-law) encodes at 8-bits to produce a bit stream of 64 kbps. During the early days of internet, this bit rate was too high to use and the available bandwidths necessitated the use of compression techniques. However, with the advent of Broadband Access Networks, it is possible to carry high bandwidth data. It is now possible to use 16-bit depth in wideband codecs as the resulting high bitrates can still be carried comfortably over a broadband link. The use of 16-bit depth reduces the theoretical SQNR for a pure sine-wave to 98.09[3] dB.
Sub-Band coding
Voice Quality
Applications:
Wireline
ITU-T G.711.1 ITU-T G.722 ITU-T G.722.2
Wireless
Cellular Mobile Wireless
Cordless Phone
Wideband Codecs Table
Testing Wideband Voice
PESQ Extensions to test Wideband Telephone Networks and Codecs ITU-T P.862.2
References
External Links
- TI Whitepaper: Wideband Voice Challenges
- DSP Designline Article: Tip: Wideband vs. narrowband VoIP codecs
- EURASIP Paper: G.711.1: A Wideband Extension to ITU-T G.711
- NTT Technical Journal Paper: Global Standard for Wideband Speech Coding: ITU-T G.711.1 (G.711 wideband extension)
- DECT Forum Website
- ITU-T P.862.2
- 3GPP2 specification
- Tutorial on VMR-WB
IETF RFC's related to Wideband Codecs
- RFC 4348 - Real-Time Transport Protocol (RTP) Payload Format for the Variable-Rate Multimode Wideband (VMR-WB) Audio Codec
- RFC 4424 - Real-Time Transport Protocol (RTP) Payload Format for the Variable-Rate Multimode Wideband (VMR-WB) Extension Audio Codec
- RFC 5391 - RTP Payload Format for ITU-T Recommendation G.711.1 (PCMA-WB and PCMU-WB)