It starts with a complaint from the sales team. Calls are breaking up. A client says you sound like a robot. Someone hangs up and rings back on their mobile instead. We’ve all been there a million times. Poor VoIP call quality rarely announces itself with a clear error message. It shows up as friction: missed words, one-sided audio, awkward silences, or that horrible, echoing sensation that makes a professional conversation feel unreliable.
The good news is that poor VoIP call quality is almost always diagnosable and fixable, provided you work through the problem systematically. The challenge is that a VoIP call touches many layers of your infrastructure simultaneously, from the codec negotiated between endpoints to the physical cable in the wall. Blaming just one thing without evidence wastes time and rarely solves anything.
This guide walks you through 12 concrete troubleshooting steps, in the order they are most useful, so you can isolate the cause and resolve it efficiently. Whether you are running Fuse 2 SIP trunking solutions into an on-premises PBX, using Direct Routing as a Service for Microsoft Teams, or working with a fully managed 3CX cloud phone system, the underlying diagnostic principles are the same. The FCC official guide to VoIP technology and regulations offers useful background on how VoIP works at a regulatory and architectural level if you want to ground yourself before diving in.
1. Reproduce the problem consistently
Before you touch a single setting, you need to understand exactly when and where the problem occurs. Asking the people experiencing it to be specific goes a long way.
- Is it every call, or only calls to certain destinations?
- Does it happen at particular times of day?
- Is it inbound audio, outbound audio, or both?
- Is the problem present from day one of a new deployment, or has it developed over time?
Reproducibility is your most important diagnostic asset. Typically, a problem that happens every time under specific conditions is far easier to isolate than one that appears randomly. Document what you know before you start changing anything, because untested changes in a live telephony environment can introduce new problems while obscuring the original one.
2. Check your bandwidth and utilisation
VoIP is sensitive to your network limits and performance. A single call using the G.711 codec (the standard used on most PSTN-quality SIP calls) requires approximately 87 kbps in each direction. Multiply that by the number of concurrent calls your business makes at peak times, then add overhead for your other traffic. Many organisations underestimate this, particularly after switching from ISDN lines where capacity was fixed and obvious.
We strongly advise running bandwidth tests during the peak hours, not during quiet times when congestion is absent. If your connection is at or near capacity during busy periods, VoIP packets will queue behind bulk data transfers and arrive late or not at all. Consider the following when assessing your connection:
- Total contracted bandwidth versus actual sustained throughput at peak
- Number of concurrent VoIP calls at your busiest hour
- Other bandwidth-intensive applications running simultaneously, such as cloud backups, video conferencing, or large file transfers
- Whether your internet connection is shared across sites or dedicated to a single location
3. Measure the core VoIP quality metrics
There are three metrics that define whether a network can support good VoIP:
- Latency – the one-way delay between sender and receiver; for VoIP, anything above 150 ms one-way begins to make conversations feel unnatural.
- Jitter – Jitter is the variation in that delay from packet to packet, and it causes audio to sound choppy or robotic.
- Packet loss – Packet loss is exactly what it sounds like: packets that never arrive, producing gaps or clipping in the audio stream.
The IETF RFC 4710: Real-time VoIP QoS monitoring framework defines the formal standards for measuring these metrics end-to-end. In practical terms, you can use a VoIP testing tool or a packet analyser such as Wireshark to capture these values during a live or simulated call. The table below summarises acceptable thresholds:
| Metric | Acceptable threshold | Likely impact if exceeded |
| One-way latency | Below 150 ms | Conversational overlap, awkward pauses |
| Jitter | Below 30 ms | Choppy or robotic audio |
| Packet loss | Below 1% | Clipping, gaps, dropped syllables |
| MOS score | 4.0 or above | Below 3.5 indicates a noticeable quality issue |
Mean Opinion Score (MOS) is a standardised measure of perceived call quality on a scale of 1 to 5. For a thorough explanation of how MOS scores are calculated alongside ITU measurement standards, the VoIP call quality metrics: MOS scores, ITU standards, and measurement methods reference from VoIP-Info.org is an excellent resource. Peer-reviewed empirical research, including the IEEE peer-reviewed study on VoIP QoS: packet loss, jitter, delay, and MOS scores, confirms that even modest packet loss has a disproportionately negative impact on perceived quality compared to equivalent degradation in other protocols.
4. Identify where in the path the problem lives
Once you have confirmed that one or more of those metrics is out of acceptable range, the next question is: where?
A VoIP packet travels from the endpoint (handset or softphone) through your local network, through your router, across your ISP’s infrastructure, and on to the carrier or cloud platform handling the call. The problem could be anywhere along that path, and it is important not to assume it is in your own environment.
A traceroute from your network to your SIP provider’s endpoint will show you hop-by-hop latency. Run it during a period of poor call quality, but not after the fact. For a concrete, illustrated example of how to read these results and identify whether the problem lies inside your network or further upstream with your ISP, see this real-world VoIP troubleshooting walkthrough using network path analysis from PingPlotter, which annotates actual latency and packet-loss graphs to show exactly where congestion is occurring. If the degradation starts at a hop outside your control, that points squarely at your ISP or your provider’s network rather than your internal configuration.
Tools to use at this stage
- Traceroute / tracert: maps hop-by-hop latency to your SIP server or carrier endpoint
- Ping with extended duration: reveals intermittent loss that a single ping will miss
- Wireshark: captures RTP (Real-time Transport Protocol) streams to measure jitter and loss at the packet level
- Your provider’s monitoring portal: if your provider offers real-time call data, use it alongside local diagnostics
5. Review your QoS configuration
Quality of Service (QoS) is the mechanism by which your router and switches prioritise certain traffic over others. Without QoS, a large file download or a software update will compete with a live VoIP call for the same bandwidth, and VoIP will often lose. QoS allows you to tag VoIP packets with a high-priority marking (typically using DSCP, Differentiated Services Code Point) so that your network equipment forwards them ahead of less time-sensitive data.
Many organisations deploy VoIP without ever configuring QoS, or configure it only on their router while leaving managed switches set to defaults. Check the following:
- Is DSCP marking enabled on your router’s WAN interface for SIP and RTP traffic?
- Do your managed switches honour DSCP markings, or do they strip them?
- Is your ISP connection QoS-aware, or does it flatten all traffic at the access point?
- Are SIP signalling ports (typically UDP 5060 or TCP 5061 for encrypted SIP) and RTP media ports included in your QoS policy?
A good router or unified threat management appliance will let you create a dedicated traffic class for VoIP and guarantee it a minimum share of your available bandwidth, preventing voice being squeezed out during busy periods.
6. Investigate your local network hardware
Consumer-grade or ageing hardware is a frequent cause of poor VoIP call quality that sneaks under the radar. Your router could be working perfectly healthy, and you could still experience latency and jitter just because your router is underpowered for the amount of simultaneous sessions it’s handling. The same applies to unmanaged switches, which cannot prioritise traffic at all, and to wireless access points operating in congested frequency bands.
Routers and firewalls
Check your router’s CPU and memory utilisation under load. Some firewalls also perform deep packet inspection on SIP traffic in a way that interferes with call setup and media, a phenomenon known as SIP ALG (Application Layer Gateway). SIP ALG is intended to help VoIP traverse NAT (Network Address Translation), but in practice it frequently mangles SIP headers and causes one-way audio or dropped calls. Disable SIP ALG on your router if it is enabled, and test again.
Switches
Unmanaged switches cannot prioritise VoIP traffic. If your IP phones share switch ports with PCs and other high-bandwidth devices, consider placing them on a dedicated VLAN (Virtual Local Area Network) using a managed switch. A voice VLAN isolates VoIP traffic from the rest of your data network and makes QoS configuration both simpler and more effective.
Wi-Fi
Wireless connections introduce variable latency and are susceptible to interference from neighbouring networks, microwave ovens, and other 2.4 GHz devices. If staff are using softphones or mobile VoIP clients over Wi-Fi and experiencing quality issues, try testing the same call over a wired connection. If the wired call is clean, the problem is in your wireless infrastructure, not your SIP trunk or internet connection.
7. Examine your codec selection
A codec is the algorithm that compresses and decompresses audio for transmission across your network. The codec negotiated between two endpoints directly affects both the bandwidth consumed by each call and the audio quality delivered. The two most common codecs in business VoIP are G.711 and G.729.
| Codec | Bandwidth per call | Audio quality | Compression |
| G.711 | ~87 kbps | Excellent (PSTN-equivalent) | None (uncompressed) |
| G.729 | ~31 kbps | Good (slight reduction) | High |
| G.722 | ~87 kbps | HD audio (wideband) | None (wideband) |
| Opus | Variable (6–510 kbps) | Excellent at medium bitrates | Adaptive |
If your bandwidth is constrained, G.729 reduces the per-call footprint significantly. However, if your network is healthy and your calls are still sounding poor, switching from G.729 to G.711 can improve perceived quality because there is no compression artefact. Check what codec your PBX or UCaaS (Unified Communications as a Service) platform is negotiating and whether it matches what your SIP provider supports at the far end. A codec mismatch forces a transcoding step, which adds latency and degrades quality. For background on how codec negotiation fits into the broader VoIP protocol stack, the VoIP protocols and quality standards overview from Washington University covers this in useful academic depth.
8. Check your SIP trunk configuration
If your metrics are within acceptable ranges and your network hardware is sound, the problem may lie in how your SIP trunk is configured. SIP trunking (Session Initiation Protocol trunking) is the technology that carries your calls between your on-premises or cloud PBX and the public telephone network. Misconfiguration here accounts for a significant proportion of call quality complaints that initially appear to be network problems.
Common SIP configuration issues
- Incorrect RTP port ranges: if your firewall is blocking or rate-limiting the UDP ports used for media, audio will be one-way or absent entirely
- NAT traversal problems: if your PBX is sending its private IP address in SIP headers rather than its public IP, media may be routed incorrectly
- Registration timeouts: a SIP trunk that is re-registering frequently can cause brief audio outages that look like call quality issues
- DTMF signalling method mismatch: if in-band DTMF does not match what your provider expects (RFC 2833 or SIP INFO), keypad presses may not register, which is a separate issue but often reported alongside quality concerns
Fuse 2’s Orca Portal for SIP trunk and direct routing management gives you real-time visibility into your trunk registrations, call statistics, and channel usage, which makes it significantly easier to spot configuration-related anomalies without needing to dig into raw SIP logs.
9. Review your firewall and NAT settings
Firewalls and NAT are the source of more VoIP quality problems than most people expect. A firewall that performs stateful inspection on SIP traffic can rewrite headers incorrectly, drop packets that it misidentifies as malicious, or time out connections during periods of silence on a call. NAT introduces the additional challenge that your PBX’s internal IP address is not routable on the public internet, so media packets need to be correctly mapped to your public address before they leave your network.
The key settings to review include:
- Disable SIP ALG (also called SIP helper or SIP fixup depending on your vendor) if it is enabled
- Confirm that your firewall allows outbound UDP traffic on the RTP port range your PBX uses (commonly UDP 10000–20000, but check your platform’s documentation)
- Ensure your NAT configuration correctly maps your PBX’s internal address to its public IP in all SIP INVITE and SDP (Session Description Protocol) packets
- Check that your firewall’s connection tracking timeout is long enough to sustain a call; some firewalls drop UDP sessions after 30–60 seconds of inactivity, which silences calls with quiet periods
10. Assess your ISP connection and upstream path
If your internal network, hardware, and configuration are all correct, but poor VoIP call quality persists, the problem is likely upstream. Your ISP connection is the next place to look. Contended broadband connections, particularly standard FTTC (Fibre to the Cabinet) services, can experience significant congestion during peak hours as bandwidth is shared between multiple premises on the same cabinet.
You should take into account whether your current connectivity product is appropriate for your call volume. A dedicated leased line or an EFM (Ethernet in the First Mile) connection provides symmetric bandwidth with a guaranteed SLA (Service Level Agreement) and is considerably more suited to business VoIP than a consumer or low-tier business broadband product. If you are on a shared connection and experiencing peak-hour quality problems, that pattern is diagnostic in itself. Contact your ISP with the traceroute and ping data you have gathered, and request that they investigate packet loss or congestion at the access network level. Some ISPs also apply traffic shaping policies that deprioritise UDP traffic, which directly affects RTP media streams.
11. Evaluate your VoIP provider’s network
Not all SIP providers are equal. The quality of the carrier network that sits behind your SIP trunk determines what happens to your call once it leaves your infrastructure. Expect latency and packet loss that no amount of internal configuration will compensate for if your provider is operating on a poorly peered or oversubscribed network
When evaluating provider-side quality, ask:
- Does your provider operate its own carrier-grade infrastructure, or does it resell capacity from a wholesale platform?
- Does the provider use an SBC (Session Border Controller), a dedicated device that manages SIP signalling at the network edge and protects against malformed packets, toll fraud, and quality degradation?
- How many points of presence (PoPs) does the provider have, and are any of them geographically close to your sites and to the destinations you call most frequently?
- Is 24/7 UK-based technical support available if a problem arises outside business hours?
These questions matter because your call quality is only as good as the weakest point in the end-to-end path. You can explore Fuse 2’s global voice and data network infrastructure to understand how our carrier-grade SBC network and multiple PoPs are designed to maintain consistent quality across international destinations. We also operate as an independent cloud communications provider, which means we are not dependent on a single upstream wholesale carrier and can route traffic intelligently to maintain quality.
12. Establish ongoing monitoring so you catch problems early
Troubleshooting after users complain is reactive. The most effective organisations move to proactive monitoring, where quality degradation is flagged before it affects enough calls to generate complaints. This means instrumenting your environment to capture the key metrics discussed in step 3 on a continuous basis, not just during incidents.
Practical monitoring for VoIP quality includes:
- SNMP monitoring on your router and switches to alert on bandwidth saturation
- SIP-specific monitoring tools that track registration status, call setup success rates, and RTP stream quality in real time
- Regular synthetic call tests, where a test call is placed on a schedule and its quality metrics recorded automatically
- CDR (Call Detail Record) analysis to identify patterns such as calls to specific destinations consistently showing lower MOS scores
If your provider offers a management portal with built-in call quality reporting, use it! Fuse 2 customer portals for billing and call management give you access to call records and usage data that complement your internal monitoring and make provider-side investigations faster to resolve. A well-monitored VoIP environment turns most quality issues from crises into minor changes.
Frequently asked questions about VoIP call quality
What causes poor VoIP call quality?
The most common causes are:
- Insufficient bandwidth,
- High network latency
- Jitter (inconsistent packet delivery timing),
- Packet loss, incorrect QoS configuration,
- Outdated or misconfigured hardware such as routers and SIP handsets.
Provider-side issues, which include poorly peered carrier networks or overloaded infrastructure, can also be a factor even when your internal environment is correctly configured.
What is an acceptable jitter level for VoIP?
Jitter below 30 milliseconds is generally considered acceptable for VoIP. Above that threshold, callers will begin to notice choppy or robotic-sounding audio. A jitter buffer, built into most modern IP-PBX platforms and routers, can absorb minor fluctuations, but it cannot compensate for sustained high jitter caused by network congestion or poor routing.
How much bandwidth does a VoIP call need?
A single VoIP call using the G.711 codec requires approximately 87 kbps of bandwidth in each direction. Lower-bandwidth codecs such as G.729 reduce this to around 31 kbps per call, though with a modest trade-off in audio fidelity. Always calculate your peak concurrent call volume when sizing your internet connection, and add a buffer for other business traffic.
What is a MOS score in VoIP?
MOS stands for Mean Opinion Score. It is a standardised measure of perceived call quality on a scale of 1 (bad) to 5 (excellent). A score of 4.0 or above is considered good for business telephony. Scores below 3.5 indicate a problem that users are likely to notice and report. Most modern VoIP platforms and monitoring tools can calculate and report MOS scores automatically from captured RTP stream data.
Can my ISP cause VoIP call quality problems?
Yes, and more often than most people realise! ISP-side packet loss, congestion at peering points, or traffic shaping policies that deprioritise real-time audio (UDP traffic) can all degrade VoIP quality even when your internal network is well configured. Running a traceroute and extended ping test to your SIP provider’s endpoint during a period of poor quality will reveal whether degradation starts inside your network or further upstream.
Not sure where your quality problem is coming from? We can help you find it
If you’re still sustaining issues after accurately following these steps, or if you would simply rather have an expert review your setup, talk to us about Fuse 2 SIP trunking solutions and how our experienced UK engineering team can assess your environment, identify the root cause, and recommend the right fix. You can also explore the Fuse 2 blog for VoIP, SIP trunking and cloud communications insights to go deeper on any of the topics covered here.
