Mean Opinion Score (MOS) is a numerical measure used to assess the quality of voice transmissions in telecommunications networks. Originally developed as a subjective rating system where human listeners would evaluate call quality on a scale from 1 (poor) to 5 (excellent), MOS has evolved into a standardized benchmark that helps service providers monitor and maintain voice quality across their networks.
How Mean Opinion Score Works
In its traditional form, MOS relies on human testers sitting in controlled environments who listen to voice samples and rate the quality based on their subjective perception. The International Telecommunication Union (ITU) standardized this process in recommendation P.800, specifying testing conditions such as room size, noise levels, and evaluation procedures.
Modern MOS ratings are calculated using various network performance metrics that predict how users would rate the call quality. These objective methods simulate human perception without requiring actual human testers, making continuous monitoring possible.
The standard MOS scale follows the Absolute Category Rating (ACR):
- 5 (Excellent): No perceptible issues
- 4 (Good): Slight imperfections that don't affect usability
- 3 (Fair): Noticeable issues but still usable
- 2 (Poor): Significant quality issues that affect usability
- 1 (Bad): Call quality is unacceptable
In practice, most VoIP systems aim for a MOS of 4.0 or higher, with scores below 3.5 generally considered problematic for business communications.
Factors Affecting MOS
Several technical factors directly impact MOS ratings in VoIP and cloud communications:
Latency: The delay between when speech is transmitted and received. High latency creates awkward conversations where parties talk over each other. For good call quality, one-way latency should remain below 150ms.
Jitter: The variation in packet arrival times. When voice data packets arrive at inconsistent intervals, audio can sound choppy or distorted. Jitter buffers help minimize this issue, but excessive jitter still degrades call quality.
Packet Loss: The percentage of data packets that fail to reach their destination. Even small amounts of packet loss can significantly impact voice quality, causing words or sounds to drop out entirely.
Codec Selection: The compression/decompression algorithm used affects both bandwidth requirements and voice quality. Non-compressed codecs like G.711 generally provide better voice quality but require more bandwidth than compressed alternatives like G.729.
Network Congestion: When networks become congested with traffic, all of the above factors tend to worsen simultaneously, causing MOS scores to drop.
Why MOS Matters
For businesses using Voice over IP (VoIP) and cloud communication services, MOS provides a quantifiable way to measure and maintain call quality. Rather than relying on vague user complaints about "poor calls" or "choppy audio", MOS offers a standardized metric that helps:
- Establish baseline performance expectations
- Identify quality issues before they become customer problems
- Compare performance across different network configurations
- Verify Quality of Service (QoS) implementations
- Support Service Level Agreements (SLAs)
At Plivo, we continuously monitor MOS across our global voice network to ensure high-quality connections. Our Points of Presence (PoPs) in key regions around the world route calls with minimal latency, and our real-time data-driven routing engine prioritizes carriers based on quality parameters including latency, jitter, and MOS scores.