VoIP Call Quality Issues in Call Centers: Causes and How to Fix Them

Poor call quality is one of the most disruptive problems in an outbound call center. Agents cannot close deals when customers cannot hear them. Choppy audio causes customers to hang up. Echo makes conversations nearly impossible.

The good news: most VoIP call quality problems have identifiable causes and straightforward fixes. This guide covers the most common call quality issues in VICIdial and Asterisk-based call centers — what causes each one and exactly how to fix it.

Understanding VoIP Audio: Why It Breaks

Unlike traditional phone calls that travel over dedicated copper circuits, VoIP calls break audio into small data packets and send them over your internet connection. The call sounds good when those packets arrive in order, on time, and without loss.

When the network is stressed, packets get delayed (latency), arrive out of order (jitter), or go missing entirely (packet loss). Each of these problems produces different audio symptoms — and each has different causes.

Issue 1: Choppy or Robotic Audio

Symptoms: Audio breaks up intermittently. Voice sounds robotic or digitized. Words cut in and out.

Cause: Jitter — packets arriving out of order or at irregular intervals.

Measurement: Use a network test tool like PingPlotter or VoIP test at voipcalculator.com. Jitter above 30ms causes audible choppy audio.

Fixes:

  1. Enable jitter buffer on your IP phones and softphones: A jitter buffer smooths out packet timing variations. Most VoIP devices have this setting — enable it and set to "adaptive" mode.

  2. Configure QoS on your router: Quality of Service prioritizes VoIP packets over regular internet traffic. Without QoS, a file download or video stream can spike jitter enough to destroy call quality.

Router QoS → Prioritize UDP traffic → Ports 10000-20000 (RTP)
Router QoS → Prioritize UDP port 5060 (SIP)

  1. Switch to a wired connection: WiFi introduces variable latency and jitter. For call center agents, ethernet is always preferred.

  2. Check for network congestion: Run a speed test during peak calling hours. If bandwidth is saturated, calls compete with other traffic. Add bandwidth or enforce traffic shaping.

Issue 2: Echo

Symptoms: The caller hears their own voice echoed back. Can also manifest as the agent hearing themselves.

Cause: Echo occurs when audio from the speaker leaks back into the microphone. This is either acoustic echo (physical leakage) or electrical echo (line impedance mismatch).

Fixes:

  1. Use noise-cancelling headsets with close-fitting ear cups: Open-ear headphones allow speaker audio to leak into the microphone. A good headset with tight ear sealing (Jabra, Plantronics) eliminates most acoustic echo.

  2. Enable echo cancellation in Asterisk: Asterisk has built-in echo cancellation. Verify it is enabled in your chan_sip.conf or pjsip.conf:

[general]
echocancelwhenbridged=yes

  1. Lower speaker volume: High speaker volume on headsets increases acoustic echo risk. Set agent headset volume to 60–70% maximum.

  2. Check SIP trunk configuration: Echo from the carrier side often indicates an impedance mismatch in the SIP trunk configuration. Contact your carrier — many offer echo cancellation at the trunk level.

Issue 3: One-Way Audio

Symptoms: The agent can hear the customer but the customer cannot hear the agent — or vice versa.

Cause: Almost always a NAT (Network Address Translation) configuration problem. When your VICIdial server is behind a router or firewall, audio packets sent back from the carrier may not know where to go.

Fix:

  1. Set your external IP in Asterisk: Edit /etc/asterisk/sip.conf:
[general]
externip=YOUR_PUBLIC_IP_ADDRESS
localnet=192.168.0.0/255.255.0.0
  1. Open RTP ports on your firewall: Audio travels over RTP, not SIP. Make sure UDP ports 10000–20000 are open both inbound and outbound:
Firewall rule: Allow UDP 10000-20000 IN and OUT

  1. Use STUN: If your server's public IP changes (dynamic IP), configure a STUN server to detect the correct external address automatically.

  2. Enable nat=yes on SIP peers: In your SIP configuration:

[carrier_name]
nat=force_rport,comedia

Issue 4: Calls Dropping Mid-Conversation

Symptoms: Active calls disconnect without warning. Agents report calls ending abruptly at random intervals (often around 30 or 60 seconds).

Cause 1 — SIP timeout: If no audio is detected for a configured period, some SIP devices assume the call ended and send a BYE. Calls that drop at exactly 30 seconds are usually a SIP timeout issue.

Fix: Enable SIP keep-alive:

[general]
qualify=yes
qualifyfreq=30

Cause 2 — RTP timeout: Asterisk has an RTP timeout setting that drops calls if no audio packets are received. For calls where AMD holds the audio briefly, this can trigger incorrectly.

Fix: In rtp.conf:

[general]
rtptimeout=60
rtpholdtimeout=300

Cause 3 — Packet loss: High packet loss causes the SIP session to fail. Check packet loss with a continuous ping test during calling hours. Packet loss above 3% causes noticeable problems; above 5% causes dropped calls.

Issue 5: High Latency / Delay in Conversation

Symptoms: Noticeable delay between when someone speaks and when the other party hears it. Conversations feel awkward, parties talk over each other.

Cause: Network latency — the time it takes packets to travel from your server to the SIP carrier and back.

Measurement: Target round-trip latency under 150ms. Above 200ms causes noticeable conversation delays.

Fixes:

  1. Choose a SIP carrier with servers close to your location: If your call center is in the Philippines, use a carrier with an Asia-Pacific POP (Point of Presence) — not one that routes through the USA.

  2. Host your VICIdial server in a data center close to your carrier: A cloud VPS in Singapore performs much better for Philippines-based calls than one in Virginia.

  3. Use G.711 codec instead of G.729: G.729 uses compression that adds processing latency. G.711 (ulaw/alaw) has zero compression latency — at the cost of higher bandwidth.

Issue 6: Audio Quality Degrades at Peak Hours

Symptoms: Call quality is good in the morning but degrades when all agents are working simultaneously.

Cause: Bandwidth saturation or server CPU overload under full agent load.

Fixes:

  1. Check bandwidth utilization during peak hours: Each concurrent call uses roughly 80–100 Kbps. For 20 agents all on calls: 2 Mbps minimum. Add 50% buffer for overhead.

  2. Check Asterisk server CPU: Run top on your VICIdial server during peak hours. If CPU usage is above 80%, the server is processing audio too slowly.

top -bn1 | head -20
  1. Check concurrent channel count: Verify your SIP trunk plan supports enough concurrent channels for your agent count. If you have 20 agents but only 15 concurrent channels, calls will fail or audio will degrade.

Issue 7: AMD Misclassification Due to Audio Quality

Symptoms: AMD false positive rate spikes. NOTSURE classifications increase. AMD accuracy becomes inconsistent.

Cause: Poor audio quality affects AMD accuracy. When audio arrives distorted, clipped, or with significant packet loss, the AMD engine cannot reliably analyze speech patterns.

This is a hidden consequence of call quality problems — not just a nuisance for agents, but a direct cause of dropped live calls and revenue loss.

Fixes:

  • Fix the underlying audio quality issues above
  • Ensure AMD is analyzing clean audio — not audio corrupted by jitter or packet loss
  • Consider AI-powered AMD (amdify.io) which is more resilient to audio quality variations than rules-based Asterisk AMD

See best AMD dialer settings for AMD configuration that handles imperfect audio better.

Network Testing Checklist

Run these tests before going live and weekly during operation:

  • Ping your SIP carrier's server — latency under 100ms?
  • Continuous ping test for 60 minutes — any packet loss?
  • PingPlotter test — jitter under 20ms?
  • Bandwidth test at peak hours — enough headroom?
  • Firewall — UDP 5060 and 10000–20000 open both directions?
  • QoS configured on router?
  • All agents on wired ethernet (not WiFi)?
  • External IP configured in Asterisk sip.conf?

When to Call Your SIP Carrier

Some call quality problems originate on the carrier side, not yours:

  • Echo that appears immediately when the call connects (before any agent interaction)
  • Audio quality that varies by the destination number being called
  • Problems that only occur on calls to specific countries or area codes
  • Quality that is consistent regardless of your network conditions

Document the call recordings, timestamps, and called numbers when reporting to your carrier. Good carriers can trace routing issues on their end and resolve them quickly.