Alcatel-Lucent algorithm is used in new ATIS industry standard supporting voice quality over next generation networks

Alcatel-Lucent (Euronext Paris and NYSE: ALU) today announced that ANIQUE+ (Auditory Non-Intrusive QUality Estimation Plus), based on an Alcatel-Lucent algorithm, is the model found in a new ATIS industry standard that supports the monitoring of voice quality over next generation networks.

This new ATIS standard (ANSI ATIS 0100005-2006) addresses a major challenge to the telecommunications industry – how to implement a non-intrusive speech quality test that can be done in live networks in real time without requiring human listeners and that will not interrupt actual calls.

“The Alcatel-Lucent approach to this was rather unique. First we simulated the functional role of the human auditory system – how the brain collects and processes sound to extract information relevant to subjective quality ratings. We then built a statistical model to map this information to a quality scale,” said Jorge Rodriguez, Director of wireless performance & RF core support in Alcatel-Lucent’s wireless activities. “Through this perceptual and statistical modelling we were able to generate an algorithm that combined many of the auditory factors and parameters into a single objective measurement that closely mirrors the human auditory process.”

ANIQUE+, which is machine based, overcomes the challenges present in today’s subjective approaches to monitoring speech quality where people are used to evaluate speech quality of telephone networks and codecs by listening to audio signals and using tools such as the Mean Opinion Score (MOS) to define the network’s audio quality. Subjective measurements of this sort are not viable for today’s real-time, heterogeneous networks due to costs and scalability.

”ANIQUE+ is superior to “intrusive machine-based objective models” that electronically measure audio quality over networks,” explains Doh-Suk Kim, the inventor of the algorithm. “These models, because they are intrusive in nature, cannot be used for real-time monitoring of live voice traffic because they operate by comparing resulting signals over the network with samples of original signal – something that cannot be done in live networks where the traffic comes from various sources operating on different technologies.”