Skip to content
HomeSIP Network Operators Conferences (SIPNOC)SIPNOC 2017General Session - Day Two7. Determining Root Cause of Voice Defects Using Different Measurement Techniques For MOS

7. Determining Root Cause of Voice Defects Using Different Measurement Techniques For MOS

[featured_image]
Download
Download is available until [expire_date]
  • Version
  • Download 85
  • File Size 1.64 MB
  • File Count 1
  • Create Date December 4, 2017
  • Last Updated December 4, 2017

7. Determining Root Cause of Voice Defects Using Different Measurement Techniques For MOS

Presented by Richard Jobson, President, Teraquant Corporation.

In a recent survey of consumers calling contact centers, * 79% said they experienced technical problems. A significant proportion experienced voice quality problems and 68% of those said it was bad enough for them to hang up. Nearly half of consumers that experienced voice quality problems felt that poor voice quality was a sign that companies didn’t really care.

Voice quality problems impact user experience, impair productivity and cause lost customers and agent churn. High definition voice when received clearly, on the other hand, enhances understanding, recognition and increases productivity. Voice quality problems have many different root causes. These include problems relating to the multimedia codec itself, the implementation of the codec on the endpoint/VoIP phone, wireless transmission problems, packet loss and jitter on the IP network itself, and problems with transcoding if used in the network path.

This presentation will equip the attendee with listening skills to shortlist potential causes of speech defects in audio recordings and also the methodology to drill down on root cause. The paper discusses standard measurement techniques [such as ‘E’ model/R Factor; PESQ and POLQA] leading to an estimate or close correlation to the subjective Mean Opinion Score [MOS].

Live audible speech files will be replayed to illustrate each category of defect and the root cause demonstrated. examples will include one-way audio; Echo; Tunnel Voice; Choppy Voice; Clipped Voice (caused by voice activation detection/VAD issues); Synthetic Voice; Underwater Voice; choked Voice caused by starved CPU & RAM on a softphone or Starved DSP resources on SBC doing transcoding and a loss of codec synchronization producing, loud squawks.

Although these audio emanations may provoke uncontrollable chuckles when heard in a test environment, they cause significant customer dissatisfaction in operations. Attendees will be empowered to flush out all such problems in their network, troubleshooting to root cause so that such problems never happen again.

* Ref: Professor Morris Pentel, chairman at Customer Experience Foundation.

Upcoming Events