EEbeat

Traditional telecommunications technology has been limited to a sound spectrum much narrower than the range of the human voice.  However, with advancements in digital signal processing (DSP) technology, narrowband telephone voice is slowly starting to be phased out for a richer wideband audio experience based on Voice over IP technology, referred to in many circles as High Definition (HD) voice.  Offering wideband voice is one way to offer users a better experience, by offering clearer and higher quality voice. 

Wideband voice is enabled by advanced DSP technology that can capture and transmit very high quality sound at twice the frequencies of narrowband voice. Today, a typical digital telephone call mimics the lower quality of analog telecommunications technology by sampling an analog signal at rate of 8,000 samples per second. This is converted into sound in the range of approximately 200 Hz to 3.3 KHz. Wideband voice is based on 16,000 samples per second, which is double the rate of narrowband voice. The greater sampling rate capabilities of a wideband voice channel accounts for sound quality that is significantly richer and extended at both ends of the spectrum.

Since wideband codecs offer double the sample rate of narrowband codecs — providing an effective pass-band of 50 to 7,000 Hz — they simply allow for superior voice quality.
Several wideband codecs have been developed and are in use. The first wideband codec, G.722, was developed almost two decades ago and simply uses a faster version of Adaptive Digital Pulse Code Modulation (ADPCM). 

When moving to wideband voice, there are a couple of trade-offs to keep in mind.  G.722 is a simple wideband codec that consumes little DSP horsepower, but that comes at the expense of using more bandwidth on the network.  Designers can alleviate that problem by using a compression codec such as G.722.2, which leverages Wideband-Adaptive Multi-rate (WB-AMR), and compresses the bandwidth and uses of 4-5 times less.  However, using a compression codec requires more DSP horsepower to do the hard work of compression.  The best solution?  Pick a system with a DSP, and take advantage of the best of both worlds: get both wideband voice and compression benefits by using a codec such as G.722.2.

Below are a few of the technical details concerning wideband and narrowband codecs.

Wideband:
 G.722: Offers wideband voice with low DSP processing requirements similar to standard G.726 ADPCM.  However, it requires a bit-rate of 48-64 kbps per channel.
 G.722.2 (AMR-WB): Offers wideband voice and compresses the bandwidth down to 12.65 kbps per channel.

Narrowband:
 G.711, G.726: Offers narrowband voice with low processing requirements, and with typical bit-rates of 32 to 64 kbps.
 Other narrowband codecs with compression such as G.729AB and G.723.1 offer bit-rates as low as 5.3 kbps, but require higher DSP processing horsepower.

By Brent Lorenz,
Texas Instruments,  Communications Infrastructure & Voice Group

About the Author
Brent Lorenz is the IP Phone Product  Manager Communications Infrastructure and Voice Business Texas Instruments. He is responsible for developing and implementing the product definition and strategy and for TI’s wired and wireless IP phone products. Lorenz coordinates worldwide marketing, sales and business development activities in support of TI’s IP phone customers in the enterprise and residential telephony space.Previously, Lorenz held a variety of business development and sales positions at companies including Texas Instruments, IBM-Rational, Wind River, as well VoIP software startups. He has extensive experience developing and maintaining customer relationships in the digital signal processor, VoIP software, and embedded software tools areas. Lorenz received a bachelor’s degree in electrical engineering from Kansas State University and an MBA from the University of Maryland.