How to Estimate VoIP Call MOS (Objective Method): A Step-by-Step Guide

Are you tired of dealing with poor quality VoIP calls that leave your customers frustrated and unhappy? Measuring the quality of your VoIP calls is crucial to ensure an excellent customer experience. One way to do this is by calculating the Mean Opinion Score (MOS), which provides a numerical representation of the voice quality. In this article, we’ll explore the objective method of estimating VoIP call MOS, providing you with a comprehensive guide to get you started.

What is MOS?

Before we dive into the estimation process, let’s quickly understand what MOS is. MOS is a subjective measure of voice quality, ranging from 1 (worst) to 5 (best). It’s based on human perception, where a group of listeners rate the quality of a voice signal. However, this method has its limitations, as it’s time-consuming, expensive, and relies on human judgment.

That’s where the objective method comes in – it uses algorithms to estimate MOS based on various parameters, making it faster and more cost-effective. In this article, we’ll focus on the objective method, which is widely used in VoIP quality monitoring.

Objective Method of Estimating MOS

The objective method involves analyzing the audio signal to estimate the MOS score. This method is based on the ITU-T P.564 recommendation, which defines a set of algorithms to estimate MOS. The process involves the following steps:

1. Signal Acquisition: The first step is to acquire the audio signal, which can be done using various tools, such as Wireshark or Tcpdump, to capture the RTP packets.
2. Signal Processing: The captured signal is then processed to remove any silence, noise, or other unrelated audio.
3. Feature Extraction: The processed signal is then analyzed to extract various features, such as packet loss, jitter, and delay.
4. MOS Calculation: The extracted features are then fed into an algorithm, which estimates the MOS score based on the E-model (recommended by ITU-T).

Signal Acquisition

In this step, you need to capture the audio signal, which can be done using various tools, such as:

• Wireshark: A popular network protocol analyzer that can capture RTP packets.
• Tcpdump: A command-line tool that captures network traffic, including RTP packets.
• Voice Quality Analyzer Tools: Specialized tools, such as VQmon or Polystar, that can capture and analyze audio signals.

Make sure to set up the capture tool to capture the RTP packets for the desired VoIP call. The captured packets will be used for further analysis.

Signal Processing

The captured signal is then processed to remove any unwanted audio, such as silence, noise, or other unrelated audio. This step is crucial, as it helps improve the accuracy of the MOS estimation.

The signal processing involves:

• Silence Detection: Identifying and removing silent periods in the audio signal.
• Noise Reduction: Reducing background noise and other unwanted audio.
• Audio Alignment: Aligning the audio signal to ensure proper synchronization.

You can use various signal processing libraries, such as Librosa or PyAudio, to perform these tasks.

Feature Extraction

In this step, you’ll extract various features from the processed audio signal, which will be used to estimate the MOS score. The features typically include:

• Packet Loss: The percentage of lost packets during the call.
• Jitter: The variation in packet delay.
• : The average delay between the sender and receiver.
• MOS-related Features: Other features, such as signal-to-noise ratio, frequency response, and distortion, that affect voice quality.

You can use libraries, such as Pystats or SciPy, to extract these features from the audio signal.

MOS Calculation

In the final step, you’ll use the extracted features to estimate the MOS score using the E-model. The E-model is a mathematical model that takes into account various parameters, such as packet loss, jitter, and delay, to estimate the MOS score.

The E-model is represented by the following equation:

MOS = 1 + Ie - (Ie - Is) * (1 - exp(-(Ppkt + Pdelay + Ploss)/15))

Where:

• Ie: The listening effort, ranging from 1 (best) to 5 (worst).
• Is: The signal-to-noise ratio, ranging from 1 (best) to 5 (worst).
• Ppkt: The packet loss probability, ranging from 0 (no loss) to 1 (total loss).
• Pdelay: The delay probability, ranging from 0 (no delay) to 1 (total delay).
• Ploss: The loss probability, ranging from 0 (no loss) to 1 (total loss).

By plugging in the extracted features into the E-model equation, you’ll get an estimated MOS score, which can help you evaluate the quality of your VoIP calls.

Example MOS Estimation using Python

To give you a better understanding of the objective method, let’s implement a simple MOS estimation script using Python. We’ll use the Librosa library for signal processing and the SciPy library for feature extraction.

import librosa
import numpy as np
from scipy.stats import skew

# Load the audio signal
audio, sr = librosa.load('audio_file.wav')

# Remove silence and noise
audio, _ = librosa.effects.trim_silence(audio, top_db=30)

# Extract features
pkt_loss = 0.05  # packet loss probability
jitter = 10  # jitter in ms
delay = 20  # delay in ms
snr = 20  # signal-to-noise ratio
skewness = skew(audio)  # skewness of the audio signal

# Calculate MOS using the E-model
Ie = 1  # listening effort
Is = 4  # signal-to-noise ratio
Ppkt = pkt_loss
Pdelay = delay / 100
Ploss = pkt_loss

MOS = 1 + Ie - (Ie - Is) * (1 - np.exp(-(Ppkt + Pdelay + Ploss)/15))

print("Estimated MOS score: ", MOS)

Conclusion

Estimating VoIP call MOS using the objective method provides a cost-effective and efficient way to evaluate voice quality. By following the steps outlined in this article, you can estimate the MOS score and identify areas for improvement. Remember to use the E-model and extract relevant features from the audio signal to get an accurate MOS score. With this knowledge, you can optimize your VoIP infrastructure and provide an exceptional customer experience.

Don’t forget to regularly monitor and analyze your VoIP calls to ensure high-quality voice communications. By doing so, you’ll be able to identify and fix issues proactively, leading to increased customer satisfaction and loyalty.

MOS Score	Voice Quality
4.0 – 4.5	Excellent
3.5 – 3.9	Good
3.0 – 3.4	Fair
2.5 – 2.9	Poor
< 2.5	Bad

Use the above table to interpret the estimated MOS score and evaluate the voice quality of your VoIP calls.

Frequently Asked Question

Are you struggling to estimate VoIP call MOS using the objective method? Look no further! Here are the answers to your most pressing questions.