Understanding VoIP Codecs and Audio Quality

Every VoIP call you make — whether it’s a quick check-in with a colleague or a high-stakes client pitch — relies on a piece of technology most people never think about: the codec. Short for coder-decoder, a codec is the engine that compresses your voice into digital packets, sends them across the internet, and reconstructs them at the other end. The codec your system uses determines how natural the call sounds, how much bandwidth it consumes, and how resilient it is to network imperfections.

For UK businesses that have migrated (or are migrating) to VoIP ahead of BT’s PSTN switch-off in January 2027, understanding codecs is no longer optional. Choosing the wrong codec can mean muffled speech, robotic artefacts, dropped syllables, and frustrated customers. Choosing the right one delivers crystal-clear HD voice that rivals — or surpasses — the quality of traditional landlines, all while making efficient use of your broadband connection.

This guide explains how VoIP codecs work, compares every major codec used in UK business telephony, breaks down bandwidth and quality trade-offs, and gives you a practical framework for choosing the best codec configuration for your network. Whether you’re managing a five-person office in Birmingham or a 500-seat contact centre in London, the principles are the same.

What Is a VoIP Codec and Why Does It Matter?

A codec performs two essential jobs. First, it encodes your analogue voice signal into a compressed digital stream that can travel across an IP network. Second, it decodes the incoming digital stream back into audible sound. This encoding-decoding cycle happens in real time, typically processing 20–30 millisecond frames of audio hundreds of times per second.

The codec you use affects three critical dimensions of every call:

• Audio quality — how natural, clear, and intelligible the speech sounds to both parties
• Bandwidth consumption — how much of your internet connection each concurrent call uses
• Latency and processing overhead — how much computational work is needed to compress and decompress the audio in real time

These three factors are always in tension. A codec that delivers superb audio quality will typically use more bandwidth. A codec that is extremely bandwidth-efficient will sacrifice some audio fidelity or require more processing power. The art of VoIP engineering is finding the sweet spot for your specific network conditions, call volumes, and quality expectations.

Narrowband vs Wideband vs Super-Wideband Audio

Before diving into individual codecs, it’s essential to understand the three tiers of VoIP audio quality. The difference comes down to the frequency range — or bandwidth — of the audio signal being captured and transmitted.

Narrowband (8 kHz sampling, 300–3,400 Hz)

Narrowband codecs sample audio at 8,000 times per second and capture frequencies between 300 Hz and 3,400 Hz. This is the same frequency range used by the traditional PSTN — it’s what most people recognise as a “phone call” sound. Speech is perfectly intelligible, but it lacks the richness of natural conversation. Consonants like “s”, “f”, and “th” can be harder to distinguish, and background context is largely stripped away.

Wideband / HD Voice (16 kHz sampling, 50–7,000 Hz)

Wideband codecs double the sampling rate to 16,000 times per second and capture a much wider frequency range from 50 Hz to 7,000 Hz. The result is a dramatic improvement in clarity — voices sound fuller, more natural, and easier to understand. Consonant differentiation improves significantly, reducing misheard words and the need for repetition. This is what the industry calls HD Voice.

Super-Wideband and Fullband (32–48 kHz sampling)

The newest generation of codecs — most notably Opus — can operate at 32 kHz or even 48 kHz sampling rates, capturing frequencies up to 20,000 Hz. At this level, VoIP audio approaches the quality of FM radio or music streaming. While not yet common in standard business telephony, super-wideband is increasingly used in conferencing applications and UCaaS platforms.

Major VoIP Codecs Explained

Let’s examine each major codec used in UK business VoIP deployments, covering how it works, what it sounds like, and where it excels.

G.711 — The Universal Standard

G.711 is the oldest and most widely supported VoIP codec, standardised by the ITU-T in 1972. It uses Pulse Code Modulation (PCM) with no compression — the audio is simply sampled 8,000 times per second at 8 bits per sample, producing a constant bitrate of 64 Kbps per direction.

G.711 comes in two variants:

• G.711 A-law (PCMA) — used throughout the UK, Europe, and most of the world outside North America. This is the variant you’ll encounter in virtually every UK VoIP deployment.
• G.711 µ-law (PCMU) — used in North America and Japan. If your business makes frequent transatlantic calls, your system will need to support both variants.

Because G.711 applies no compression, it delivers excellent narrowband audio quality with a typical MOS score of 4.1–4.2. It also requires virtually zero processing power, making it ideal for systems handling hundreds of concurrent calls. The trade-off is bandwidth — at 64 Kbps per direction (roughly 87 Kbps with IP/UDP/RTP headers), G.711 uses more bandwidth per call than any other common codec.

G.729 — The Bandwidth Saver

G.729 was developed in the 1990s specifically to address G.711’s bandwidth appetite. It uses Conjugate Structure Algebraic Code Excited Linear Prediction (CS-ACELP) — a sophisticated compression algorithm that reduces the bitrate to just 8 Kbps per direction while maintaining surprisingly good audio quality.

With a typical MOS score of 3.9–4.0, G.729 sounds almost as good as G.711 to most listeners. The difference is subtle — voices sound slightly thinner and background nuances are lost — but speech remains perfectly clear and professional. Where G.729 truly shines is bandwidth efficiency: at roughly 31 Kbps per call (including headers), you can run nearly three times as many concurrent calls on the same internet connection compared to G.711.

G.729 does require a licence fee, which historically added £5–£10 per channel to system costs. However, many modern VoIP platforms include G.729 licensing in their per-user pricing, so this is rarely a separate line item in 2026. The codec also requires more processing power than G.711, though this is negligible on modern hardware.

G.722 — HD Voice for Business

G.722 is the codec that brought HD Voice to business telephony. Operating at 16 kHz sampling with a 64 Kbps bitrate, it captures audio frequencies up to 7,000 Hz — double the range of narrowband codecs. The result is a noticeably richer, clearer sound with better consonant differentiation and reduced listener fatigue during long calls.

Interestingly, G.722 uses the same bandwidth as G.711 (64 Kbps) but delivers substantially better audio quality thanks to its use of Sub-Band Adaptive Differential Pulse Code Modulation (SB-ADPCM). This makes it a straightforward upgrade for any network already provisioned for G.711 — you get better quality at no additional bandwidth cost.

G.722 achieves a typical MOS score of 4.3–4.5, and it is widely supported by modern IP phones from Yealink, Poly, Cisco, and Grandstream. If your VoIP provider supports it and your endpoints are compatible, enabling G.722 is one of the simplest ways to improve call quality.

Opus — The Modern Adaptive Codec

Opus is the newest and most technically advanced codec in common VoIP use. Developed by the IETF and released in 2012, it is open-source, royalty-free, and extraordinarily flexible. Opus can operate at bitrates from 6 Kbps to 510 Kbps, sampling rates from 8 kHz to 48 kHz, and frame sizes from 2.5 ms to 60 ms — all configurable in real time.

What makes Opus genuinely revolutionary is its adaptive bitrate capability. The codec continuously monitors network conditions and adjusts its compression level on the fly. If bandwidth is plentiful, Opus ramps up quality to deliver fullband audio that rivals music streaming. If the network becomes congested, it gracefully reduces bitrate to maintain a stable connection without dropping the call.

Opus achieves MOS scores of 4.3–4.9 depending on the bitrate, comfortably surpassing every legacy codec at equivalent bandwidth. It is the default codec for WebRTC (used by browser-based calling and most UCaaS platforms), Microsoft Teams, Zoom, and many next-generation VoIP systems. For UK businesses evaluating future-proof VoIP solutions, Opus support should be high on the requirements list.

iLBC — The Packet Loss Fighter

Internet Low Bitrate Codec (iLBC) was developed by Global IP Solutions (now part of Google) specifically for VoIP calls over unreliable networks. It operates at either 13.3 Kbps (30ms frames) or 15.2 Kbps (20ms frames), making it very bandwidth-efficient.

iLBC’s standout feature is its exceptional resilience to packet loss. While most codecs degrade significantly when packets go missing, iLBC uses independent frame encoding — each audio frame is self-contained and doesn’t depend on preceding frames. This means a lost packet affects only a tiny slice of audio rather than causing cascading distortion. At 5% packet loss, where G.711 calls become noticeably degraded, iLBC maintains acceptable quality.

The trade-off is audio quality under ideal conditions — with a MOS score of roughly 4.1, iLBC matches G.711 in narrowband clarity but cannot compete with wideband codecs like G.722 or Opus. It remains a valuable fallback codec for remote workers on unreliable connections, mobile hotspots, or international links with variable quality.

Bandwidth Consumption by Codec

One of the most practical considerations when choosing a codec is how much bandwidth each concurrent call will consume. This directly determines how many simultaneous calls your internet connection can support. The figures below include the overhead from IP, UDP, and RTP headers, which add approximately 23–40 Kbps depending on the packetisation interval.

To put this in practical terms: a standard UK business broadband connection delivering 80 Mbps download and 20 Mbps upload (a typical FTTC line) could theoretically support around 230 concurrent G.711 calls on the upload side alone. However, you should never allocate more than 70–80% of your upload bandwidth to VoIP to leave headroom for other traffic and avoid congestion. With QoS policies in place, a 20 Mbps upload connection comfortably supports 150+ simultaneous calls using G.711 or 400+ using G.729.

Understanding MOS Scores

The Mean Opinion Score (MOS) is the industry-standard metric for measuring perceived voice quality. Originally based on subjective listening tests where panels of people rated call quality on a scale of 1 to 5, MOS is now commonly estimated algorithmically using tools like PESQ (Perceptual Evaluation of Speech Quality) and POLQA (Perceptual Objective Listening Quality Analysis).

Here is what the MOS scale means in practice:

For UK business telephony, a MOS score of 4.0 or above is the minimum acceptable standard. Scores below 3.5 indicate a serious problem — callers will notice distortion, robotic effects, or difficulty understanding speech. Most well-configured business VoIP systems achieve MOS scores between 4.1 and 4.4 under normal network conditions.

Transcoding: The Hidden Quality Killer

Transcoding occurs when audio must be converted from one codec to another mid-call. For example, if your VoIP system uses G.722 internally but the receiving party’s system only supports G.711, a media gateway or session border controller must decode the G.722 audio and re-encode it as G.711 in real time.

Every transcoding step introduces three problems:

• Quality degradation — each encode-decode cycle is lossy; transcoding from a wideband codec to a narrowband one permanently discards the higher-frequency audio information
• Additional latency — the decode-reencode process adds 5–15 milliseconds per transcoding hop, which compounds if multiple gateways are involved
• Increased processing load — transcoding is computationally expensive, and busy media gateways handling hundreds of simultaneous transcodings can become bottlenecks

The golden rule of VoIP codec management is: minimise transcoding wherever possible. This means ensuring that your internal system, your SIP trunk provider, and the devices at both ends of the call all support a common codec. For UK businesses, the safest bet is ensuring G.711 A-law is always available as a fallback — it is universally supported and avoids transcoding in the vast majority of scenarios.

Codec Negotiation in SIP

When a VoIP call is initiated using SIP (Session Initiation Protocol), the two endpoints must agree on which codec to use. This negotiation happens automatically during call setup via a mechanism called the SDP offer/answer model (Session Description Protocol).

Here is how it works in practice:

• Step 1: The INVITE — the calling device sends a SIP INVITE message containing an SDP body that lists all codecs it supports, in order of preference. Each codec is identified by its RTP payload type number (e.g., 8 for G.711 A-law, 0 for G.711 µ-law, 9 for G.722, 18 for G.729).
• Step 2: The Answer — the receiving device examines the offered codec list, compares it against its own supported codecs, and responds with the highest-priority codec that both sides support.
• Step 3: Media flows — RTP audio packets begin flowing using the agreed codec. If conditions change mid-call, some systems support codec renegotiation via a SIP re-INVITE.

The critical point for IT administrators is that codec priority order matters enormously. If your system’s codec list has G.711 as the first choice and G.722 second, the system will always use G.711 when the remote end supports it — even though G.722 would deliver better quality at the same bandwidth. Configuring the optimal priority list is one of the most impactful changes you can make to your VoIP quality.

A recommended codec priority list for most UK business deployments is:

1. Opus (if both endpoints support it — best adaptive quality)
2. G.722 (wideband HD Voice at standard bandwidth)
3. G.711 A-law (universal fallback, excellent narrowband quality)
4. G.729 (bandwidth-constrained scenarios)
5. iLBC (poor network conditions fallback)

HD Voice: What It Means for Your Business

HD Voice is a marketing term that describes wideband audio in VoIP calls, typically delivered by codecs like G.722 or Opus. The benefits are not just cosmetic — they have measurable business impact:

• Reduced misunderstanding — wider frequency range makes consonants and accents clearer, reducing “sorry, can you repeat that?” moments by up to 40%
• Lower listener fatigue — call centre agents handling hundreds of calls per day report significantly less exhaustion with HD Voice, as the brain expends less effort reconstructing missing audio information
• Better conference calls — HD Voice makes it easier to distinguish between multiple speakers, a critical advantage for multi-party conference calls where narrowband audio quickly becomes muddy
• Professional impression — crystal-clear audio conveys professionalism and attention to detail, particularly important for client-facing businesses in sectors like law, finance, and consulting

To enable HD Voice on your UK VoIP system, you need three things:

1. IP phones or softphones that support G.722 or Opus — most business-grade IP phones manufactured since 2015 support G.722; Opus support is increasingly common in softphones and WebRTC-based systems
2. A VoIP provider that supports wideband codecs — your SIP trunk or hosted PBX must not strip or transcode wideband audio; ask your provider specifically about G.722 and Opus support
3. Adequate network quality — HD Voice is more sensitive to jitter and packet loss than narrowband audio; ensure your network meets the QoS thresholds outlined earlier

Choosing the Right Codec for Your Network

There is no single “best” codec — the right choice depends on your network conditions, call volumes, endpoint capabilities, and quality priorities. Here is a practical decision framework for UK businesses:

Scenario 1: Standard Office with Good Broadband

If your office has a reliable FTTC, FTTP, or leased-line connection with low latency and minimal packet loss, prioritise G.722 or Opus for HD Voice quality. Use G.711 A-law as your fallback. Bandwidth is unlikely to be a constraint — even a modest 40 Mbps symmetric connection supports well over 100 concurrent HD calls.

Scenario 2: High Call Volume Contact Centre

Contact centres with 100+ concurrent calls need to balance quality against bandwidth and processing capacity. G.722 is typically the best choice — it delivers HD Voice at the same bandwidth as G.711 and is universally supported by professional IP phones. For centres with bandwidth limitations, G.729 allows you to triple your concurrent call capacity with only a modest quality reduction.

Scenario 3: Remote and Hybrid Workers

Remote workers connecting over residential broadband, shared Wi-Fi, or mobile hotspots face variable network conditions. Opus is the ideal codec here — its adaptive bitrate automatically adjusts to available bandwidth, and its built-in packet loss concealment handles the inconsistencies of home networks. If Opus is not available, configure iLBC as a fallback for its superior packet loss resilience.

Scenario 4: International Calls

Calls routed over international SIP trunks may traverse multiple carriers and media gateways. To minimise transcoding, ensure G.711 is always in your codec list — it is the one codec virtually every system on earth supports. For calls to North America, be aware that their systems default to G.711 µ-law, so your SBC or VoIP platform needs to handle the A-law to µ-law conversion (which is trivial and lossless).

Scenario 5: Bandwidth-Constrained Sites

Branch offices or temporary sites operating on basic ADSL, satellite links, or 4G connections may have limited upload bandwidth. G.729 at 31 Kbps per call is the traditional solution, allowing 10+ concurrent calls on even a 512 Kbps upload connection. Alternatively, Opus in narrowband mode at 12–16 Kbps achieves similar bandwidth savings with better adaptive capability.

Codec Configuration Best Practices

Whether you manage your VoIP system in-house or rely on a managed service provider like Cloudswitched, these best practices will help you get the most from your codec configuration:

1. Audit Your Current Codec Usage

Most VoIP platforms provide call detail records (CDRs) that show which codec was used for each call. Review these records over a two-week period to understand your actual codec distribution. You may discover that calls are defaulting to G.711 when G.722 is available on both ends, simply because the priority list has not been optimised.

2. Standardise Endpoint Firmware

Inconsistent firmware versions across your IP phone fleet can lead to codec mismatches and unexpected transcoding. Ensure all devices are running current firmware with consistent codec configuration. Most enterprise phone management platforms (like Yealink Device Management or Poly Lens) can push configuration profiles to all devices simultaneously.

3. Test with Real Calls, Not Just Specifications

Codec specification sheets describe performance under ideal laboratory conditions. Real-world performance depends on your specific network, SIP trunk provider, and endpoint hardware. Conduct A/B testing — make the same call using different codecs and compare the quality. Use MOS monitoring tools to quantify the difference objectively.

4. Configure Silence Suppression Carefully

Silence suppression (also called Voice Activity Detection or VAD) stops transmitting audio during pauses in speech, saving significant bandwidth. However, aggressive silence suppression can clip the beginnings of sentences and create an unsettling “dead air” sensation. If you enable it, test thoroughly to ensure it does not affect conversation flow. Some codecs — particularly Opus — have built-in DTX (Discontinuous Transmission) that handles this more gracefully than legacy implementations.

5. Plan for Codec Changes as You Grow

Your codec strategy should evolve with your business. A five-person startup on basic broadband might start with G.729 to conserve bandwidth. As the business grows and upgrades to a leased line, switching to G.722 or Opus unlocks HD Voice quality. Revisit your codec configuration whenever you change internet provider, add significant users, or deploy new endpoints.

The Future of VoIP Codecs

The VoIP codec landscape continues to evolve rapidly. Several trends are shaping what UK businesses can expect in the coming years:

• Opus dominance — as WebRTC becomes the foundation for browser-based calling and UCaaS platforms expand, Opus is becoming the de facto standard. Its royalty-free licensing and adaptive capabilities make it the obvious choice for new deployments.
• AI-enhanced codecs — emerging codecs like Lyra (from Google) and EnCodec (from Meta) use neural networks to achieve astonishing compression ratios. Lyra, for instance, delivers intelligible speech at just 3 Kbps. While not yet mainstream in business telephony, these AI codecs will likely appear in commercial VoIP products within 2–3 years.
• End-to-end encryption — SRTP (Secure Real-time Transport Protocol) adds encryption overhead to voice streams. Efficient codecs like Opus minimise the bandwidth impact of encryption, making them better suited to security-conscious deployments.
• 5G and codec-agnostic networks — as 5G deployment accelerates across the UK, the bandwidth constraints that historically favoured low-bitrate codecs become less relevant. This further tilts the balance toward wideband and fullband codecs that prioritise quality over compression.

What Good VoIP Audio Costs in the UK

Understanding codecs helps you make smarter purchasing decisions. Here is what UK businesses should expect to invest for different quality tiers:

• Basic narrowband VoIP (G.711/G.729) — £6–£12 per user per month for a hosted PBX with standard desk phones. Perfectly adequate for most business calls.
• HD Voice VoIP (G.722) — £10–£18 per user per month. Requires HD-capable IP phones (£60–£150 per handset) or softphone licences.
• UCaaS with Opus and WebRTC — £15–£30 per user per month for platforms like Microsoft Teams Calling, 8x8, or RingCentral. Includes voice, video, messaging, and advanced analytics.
• Contact centre with codec optimisation — £25–£50 per agent per month for platforms with intelligent codec management, MOS monitoring, and real-time quality alerts.

Beyond the per-user costs, the most impactful investment is in your network infrastructure. A properly configured network with QoS policies, adequate bandwidth, and low latency is worth more to your call quality than the most expensive codec. Budget £500–£2,000 for a professional VoIP readiness assessment and QoS configuration of your switches, routers, and firewall.

How Cloudswitched Can Help

At Cloudswitched, we have deployed and optimised VoIP systems for hundreds of UK businesses — from small professional services firms to multi-site enterprises. Our approach to codec management is systematic:

• Network assessment — we measure latency, jitter, packet loss, and available bandwidth across your entire network, including remote worker connections, before recommending a codec strategy
• Codec optimisation — we configure codec priority lists, QoS policies, and SBC settings tailored to your specific network conditions and quality requirements
• Ongoing monitoring — our managed VoIP service includes real-time MOS monitoring with automated alerts when call quality drops below your defined threshold
• Future-proofing — we deploy systems that support modern codecs like Opus alongside legacy standards, ensuring your investment is protected as technology evolves