Welcome to the Sevana voice quality FAQ. Below you will find clear answers about Sevana, our products, and the voice and call quality standards we work with every day.
What this voice quality FAQ covers
This voice quality FAQ is grouped into six short sections: who Sevana is, our products, codecs and integration, MOS scoring, the standards we work with, and testing methods. If your question is not here, just ask us.
About Sevana
What does Sevana do?
In short, Sevana builds voice and audio quality analysis technology. Because audio quality is all we do, we focus on measuring how good a call really sounds and on showing where quality breaks down, across VoIP, VoLTE and 2G to 5G networks. You can learn more about Sevana here.
Which products does Sevana offer?
Our main products are AQuA for reference-based audio quality and PVQA for passive perceptual analysis. In addition, VQ Monitor handles real-time monitoring, Pcaptix handles PCAP analysis, and QualTest covers mobile drive testing.
Why should I choose Sevana?
Because audio quality is our only focus, we bring deep DSP and algorithm expertise that general vendors do not. Moreover, our technology is proven worldwide and integrates flexibly. When you need something built to fit, we also offer custom development.
Do you provide support after purchase?
Yes, and it comes from the engineers who built the technology. Therefore you reach people who actually understand the internals. Simply email info@sevana.biz or contact us whenever you need help.
How quickly can we get started?
Usually quite fast, because our binaries are self-contained and need little setup. Once licensing is in place, a dedicated contact then guides you through onboarding and your first results.
What are your contract terms?
Our terms are flexible. For example, you can end an agreement at any time with 30 days written notice, so both sides can plan resources sensibly.
Who are Sevana’s partners and customers?
We are proud to work with names such as Perfecto, Occam Networks, NetQPro and Gigaset. In addition, several major telecom players rely on us, although we cannot name them publicly. Many of these relationships span more than ten years.
Our products
What is PVQA?
PVQA (Passive Voice Quality Analysis) scores decoded speech the way a listener would, so no reference signal is required. As a result, it detects clipping, echo, noise and dead air, and then predicts a MOS on the ITU-T P.800 scale.
What is AQuA?
AQuA measures audio quality by comparing a test signal against a clean reference, although it also offers no-reference modes. Therefore it works well as a practical alternative to POLQA and ViSQOL for labs and automated testing.
What is VQ Monitor?
VQ Monitor is our real-time, passive monitoring server for VoIP and VoLTE networks. For every stream it reports two MOS models, namely the G.107 Network MOS and the Sevana PVQA MOS, and it also raises alarms and serves a live dashboard.
What is Pcaptix?
Pcaptix analyses VoLTE and VoIP call quality straight from
.pcap captures. First it extracts SIP and RTP and decodes the audio. Then it shows dual MOS, quality timelines and plain-English AI explanations.What is QualTest?
QualTest is a mobile-to-mobile drive-test probe for field testing. Because it maps network KPIs to audio quality, you can see exactly where and why quality drops on real devices.
Codecs, networks and integration
What is the difference between AQuA and PVQA?
AQuA is reference-based, so it compares a test signal with a known reference, which is ideal in a lab. PVQA, however, is reference-free, so it scores live or recorded speech with no reference. Therefore PVQA is what you need on production traffic.
Which codecs do you support?
We decode the common telephony and VoLTE codecs. For example, the list includes G.711 (A-law and mu-law), G.722, G.729, Opus, iLBC, GSM, AMR-NB, AMR-WB and EVS. In addition, AMR and EVS data carried in IuUP frames is parsed automatically.
Do you support VoLTE and 5G?
Yes, because our engine is already proven across VoIP, 2G, 3G, 4G, VoLTE and live 5G networks. Moreover, since we decode AMR-WB and EVS, perceptual scoring works on real VoLTE media, not only on legacy codecs.
Can you analyse encrypted (SRTP) media?
Yes. In encrypted-only mode, VQ Monitor still reports network metrics without any keys. Alternatively, when SIP carries SDES keys, it decrypts SRTP per call and then runs full perceptual analysis.
Can I integrate Sevana into my own system?
Yes, because our technology ships as libraries with documented APIs. In addition, we provide custom development and integration, so you can add waveform analysis and impairment detection to your existing software.
Do you offer a free evaluation?
Yes. Simply contact us, and we will arrange an evaluation of the right tool for your use case.
MOS and quality scoring
What is a MOS score?
MOS (Mean Opinion Score) is a single number, usually from 1 to 5, where 1 is the lowest perceived quality and 5 is the highest. In short, it is the standard way to express how good a call sounds.
What is the difference between Network MOS and Sevana MOS?
Network MOS is estimated from transport statistics such as loss, jitter and delay, using the ITU-T G.107 E-model. Sevana MOS, by contrast, comes from PVQA listening to the decoded audio. Therefore the first tells you what the network delivered, while the second tells you what the caller actually heard.
What is the E-model (ITU-T G.107)?
The G.107 E-model predicts a Network MOS from packet loss, jitter and delay, and it uses per-codec impairment values from G.113. Because it needs no audio decode, it is very inexpensive and also works on encrypted streams.
Standards we work with
What is POLQA?
What is ViSQOL?
What is QoE?
Quality of Experience (QoE) measures a customer’s overall delight or annoyance with a service, such as a phone call or a stream. In short, it is a holistic view of the whole experience, much like the idea of user experience.
What is ITU-T P.564?
ITU-T P.564 defines conformance testing for VoIP transmission-quality models. Because it combines packet and acoustic analysis, it gives a more reliable MOS than the E-model alone. The real-time PVQA in VQ Monitor is an alternative.
What is ITU-T P.563?
ITU-T P.563 is a single-ended method for objective speech quality in narrow-band telephony. PVQA, however, is not limited to narrow-band, since it predicts MOS across all kinds of audio.
Testing methods
What is intrusive versus non-intrusive testing?
Intrusive testing compares a test signal with a known reference, as POLQA, ViSQOL and AQuA do. Non-intrusive testing, by contrast, scores live traffic with no reference, as PVQA does. Therefore non-intrusive testing is essential when there is no clean original to compare against.
What are monitoring agents or probes?
They are automated probes, including mobile probes, that run end-to-end and single-ended tests. In practice, they measure the QoS and QoE of communications either continuously or on a schedule.
