Voice Quality Monitoring

To measure the voice quality of calls that are sent and received within the network, Cisco Unified IP Phones use the following statistical metrics that are based on concealment events. The DSP plays concealment frames to mask frame loss in the voice packet stream.

• Concealment Ratio metrics: Shows the ratio of concealment frames over total speech frames. The phone calculates an interval conceal ratio every 3 seconds.
  
• Concealed Second metrics: Shows the number of seconds in which the DSP plays concealment frames due to lost frames. A severely “concealed second” is a second in which the DSP plays more than five percent concealment frames.
• MOS-LQK metrics: Uses a numeric score to estimate the relative voice listening quality. The Cisco Unified IP Phone calculates the mean opinion score (MOS) for listening quality (LQK) based audible concealment events due to frame loss in the preceding 8 seconds, and includes perceptual weighting factors such as codec type and frame size.

The phone uses the Cisco proprietary algorithm, Cisco Voice Transmission Quality (CVTQ) index, to produce MOS LQK scores. Depending on the MOS LQK version number, these scores might be compliant with the International Telecommunications Union (ITU) standard P.564. This standard defines evaluation methods and performance accuracy targets that predict listening quality scores based on observation of actual network impairment.

Note 

Concealment ratio and concealment seconds are primary measurements based on frame loss while MOS LQK scores project a “human-weighted” version of the same information on a scale from 5 (excellent) to 1 (bad) for measuring listening quality.

Listening quality scores (MOS LQK) relate to the clarity or sound of the received voice signal. Conversational quality scores (MOS CQ, such as G.107) include impairment factors, such as delay, that degrade the natural flow of conversation.

You can access voice quality metrics from the Cisco Unified IP Phone by using the Call Statistics screen or remotely by using Streaming Statistics.

Related Topics

Call Statistics Screen

Remote Monitoring

Voice Quality Metric Interpretation

To use the metrics for monitoring voice quality, note the typical scores under normal conditions of zero packet loss and use the metrics as a baseline for comparison.

It is important  to distinguish significant changes from random changes in metrics. Significant changes are scores that change about 0.2 MOS or greater and persist in calls that last longer than 30 seconds. Conceal Ratio changes should indicate greater than 3 percent frame loss.

MOS LQK scores can vary based on the codec that the Cisco Unified IP Phone uses.

The following codecs on the Cisco Unified IP Phones 7962G and 7942G provide these maximum MOS LQK scores under normal conditions with zero frame loss:

• G.711 gives 4.5
  
• G.722 gives 4.5
  
• G.728/iLBC gives 3.9
  
• G.729 A/AB gives 3.8
  

The following codecs on the Cisco Unified IP Phones 7961G/G-GE and 7941G/G-GE provide these maximum MOS LQK scores under normal conditions with zero frame loss:

• G.711 codec gives 4.5 score
  
• G.729A/ AB gives 3.7
  

Note

• CVTQ does not support wideband (7 kHz) speech codecs, because ITU has not defined the extension of the technique to wideband. Therefore, MOS scores that correspond to G.711 performance are reported for G.722 calls to allow basic quality monitoring, rather than not reporting an MOS score.
  
• Reporting G.711-scale MOS scores for wideband calls through the use of CVTQ allows basic quality classifications to be indicated as good/normal or bad/abnormal. Calls with high scores (approximately 4.5) indicate high quality/low packet loss, and lower scores (approximately 3.5) indicate low quality/high packet loss.
  
• Unlike MOS, the Conceal Ratio and Concealed Seconds metrics remain valid and useful for both wideband and narrowband calls.
  

A Conceal Ratio of zero indicates that the IP network is delivering frames and packets on time with no loss.

Voice Quality Troubleshooting Tips

When you observe significant and persistent changes to metrics, use the following table for general troubleshooting information:

Changes to Voice Quality Metrics

Note 

Voice quality metrics do not account for noise or distortion, only frame loss.