Amazon Connect VDI Audio Quality Fix: Complete Guide for Citrix & WorkSpaces

5 Fastest Fixes for Amazon Connect VDI audio quality problem

Amazon Connect VDI Audio Quality Fix

Enable audio redirection (WebRTC offload): Add the VDIPlatform=CITRIX or VDIPlatform=AWS_WORKSPACE query parameter to your agent workspace URL so audio bypasses the VDA server and flows directly from the thin client.
Add the Citrix registry key: Open Registry Editor on the VDA, navigate to HKEY_LOCAL_MACHINE\SOFTWARE\WOW6432Node\Citrix\WebSocketService, create the ProcessWhitelist key, and add Chrome.exe and msedge.exe. Restart CtxHdxWebSocketService.
Open UDP port 3478: Confirm outbound UDP 3478 is open on the agent’s thin client firewall to TurnNlb-*.elb.{region}.amazonaws.com. This is the most common reason media stays stuck on the server.
Check your CWA and VDA versions: Use Citrix Workspace App 2305 or later and Citrix VDA 2203 LTSR or later. Older versions simply do not support WebRTC redirection.
Verify redirection is working: Open Task Manager on the thin client and look for the HDX RTC Engine process. If it’s running, media is being redirected correctly.

⚠ Note: If you are using the Native CCP (not an Embedded or Custom CCP), audio optimization is not supported. You must use an Embedded or Custom CCP built with the Amazon Connect Streams.js and Connect-RTC.js libraries.

Why VDI Causes Audio Lag in Amazon Connect

Amazon Connect VDI audio quality fix – If your contact centre agents are working inside a Citrix or Amazon WorkSpaces environment and they are complaining about choppy audio, one-way sound, delays, or echo, you are not alone. This is one of the most common support issues for Amazon Connect deployments in virtual desktop environments, and the good news is it is almost always fixable.

The core problem is simple: VDI environments were never built for real-time audio. When an agent makes or receives a call through the Amazon Connect Contact Control Panel (CCP) running inside a Citrix session, the audio has to travel an incredibly inefficient path. It goes from the customer, to Amazon Connect in the AWS cloud, back to the Citrix server, then across the ICA protocol channel to the agent’s thin client, and then back again. Every extra hop adds latency, and in real-time voice communication, milliseconds matter in a way they simply do not for loading a webpage or a CRM record.

To put some numbers to it: under 100ms round-trip latency feels seamless. At 150–200ms you may notice a slight delay. Above 300ms conversations become uncomfortable and agents start cutting each other off. When audio travels through an unoptimized VDI path, you can easily exceed 300ms just in internal routing.

Key insight: VDI adds extra network hops between the agent endpoint and Amazon Connect. Without audio redirection, audio processing happens on the Citrix server rather than the agent’s local device — this is the root cause of most lag, echo, and quality issues.

There is another cost problem too. Processing real-time audio on a VDA server is CPU-intensive. This means your Citrix environment needs to host fewer agents per server than it otherwise would, which directly increases your cost per agent. Implementing the proper Amazon Connect VDI audio quality fix also reduces your infrastructure bill.

How Audio Optimization Actually Works

Amazon Connect’s solution to the VDI audio problem is called audio redirection, and it is elegant. Instead of sending voice traffic through the VDA server, the system offloads all audio processing to the agent’s local thin client device. The call audio travels directly between the thin client and Amazon Connect’s Softphone Media service. The Citrix VDA server only handles the signalling and the browser UI, not the heavy lifting of encoding and decoding audio in real time.

Here is the step-by-step flow when everything is set up correctly:

Call Enters Amazon Connect – A customer calls in. Amazon Connect receives the call and routes it to the available agent through the normal flow.
CCP Receives the Notification – The agent’s Contact Control Panel (CCP), running inside the Citrix browser session, is notified of the incoming call via a persistent WebSocket connection on TCP port 443.
Citrix UC SDK Takes Over – The CCP uses the Amazon Connect Connect-RTC.js library, which contains the Citrix Unified Communications (UC) SDK. This library detects it is running in a Citrix VDI environment and signals the Citrix UC Media Engine on the thin client via the ICA virtual channel.
Thin Client Establishes Direct Media Connection – The Citrix UC Media Engine running on the agent’s local device establishes a direct audio connection to Amazon Connect’s Softphone Media service over UDP port 3478. Audio now bypasses the VDA server entirely.
High-Quality Audio Flows Directly – Voice travels peer-to-peer between the agent’s thin client and Amazon Connect. Network hops are reduced, latency drops, and audio quality improves significantly.

This architecture is supported for both Citrix Virtual Desktops and Amazon WorkSpaces (both PCoIP and DCV protocol variants). Amazon also supports Omnissa (formerly VMware Horizon) using a similar approach with a different SDK.

System Requirements Before You Start

Before you troubleshoot anything, confirm that your environment actually meets the minimum requirements for audio optimization. Many support cases turn out to be version mismatches that no amount of configuration will fix.

Check – KloudMint’s Amazon Connect VDI Compatibility Tester Tool

For Citrix VDI

Component	Minimum Version	Recommended	Status
Citrix Workspace App (CWA)	2305 (Windows, Linux, Mac)	Latest available	Check first
Citrix VDA (Virtual Delivery Agent)	2203 LTSR	Latest LTSR	Check first
Browser (inside VDA session)	Chrome or Microsoft Edge	Latest Chrome	Required
CCP Type	Embedded or Custom CCP only	Custom CCP with Connect-RTC.js	Native CCP: Not supported
CtxHdxWebSocketService	Must be running on VDA	Running, not stopped	Verify in Task Manager

For Amazon WorkSpaces

Protocol	Virtual Driver Required	Action
PCoIP	Teradici Virtual Audio Driver	Install via Device Manager if not listed under Sound controllers
DCV (NICE)	AWS Virtual Speakers 7.1 Device + AWS Virtual Microphone Device	Install from `C:\Program Files\NICE\DCV\Server\drivers\audio`

Quick check: On your WorkSpace, open Device Manager and expand “Sound, video and game controllers.” If you do not see the Teradici (PCoIP) or AWS Virtual audio drivers listed, that is your first fix — install them before anything else.

Fix: Amazon Connect VDI audio quality fix on Citrix

Follow these steps in order. Each step builds on the previous one, and skipping ahead is a common mistake that leads to wasted time.

Step 1 — Add the Citrix WebSocket Allowlist Registry Key

This is the most commonly missed step. Without this registry key, the Citrix UC SDK cannot communicate with the browser running inside the VDA session, so WebRTC redirection is silently blocked.

On your Citrix VDA server (not the thin client), open the Windows Registry Editor and create the following entry:Registry Key — Citrix VDA Server

Key Path:  HKEY_LOCAL_MACHINE\SOFTWARE\WOW6432Node\Citrix\WebSocketService
Key Name:  ProcessWhitelist
Key Type:  REG_MULTI_SZ
Key Value: Chrome.exe
           msedge.exe
           Firefox.exe

After adding the registry key, you must restart the CtxHdxWebSocketService for the change to take effect. Open Task Manager → Services tab, find CtxHdxWebSocketService, right-click, and select Restart.

Important: You need to set this on the VDA (the server running your Citrix session), not on the agent’s thin client machine. Administrators often confuse the two machines when applying this fix.

Step 2 — Configure the CCP with the VDIPlatform Parameter

If you are using the Amazon Connect Agent Workspace, you need to tell it which VDI environment it is running in by appending a query parameter to the URL.Agent Workspace URL — Citrix

https://your-instance-url.my.connect.aws/agent-app-v2?VDIPlatform=CITRIX

Agent Workspace URL — Amazon WorkSpaces

https://your-instance-url.my.connect.aws/agent-app-v2?VDIPlatform=AWS_WORKSPACE

Bookmark this URL for all your agents. This is what signals the CCP to initiate WebRTC redirection through the Citrix UC SDK rather than processing audio directly through the browser on the VDA.

If you use SSO, you need to embed the VDIPlatform parameter into your Identity Provider’s relay state URL instead:SSO Relay State URL — Citrix

https://us-east-1.console.aws.amazon.com/connect/federate/instance-id
  ?destination=%2Fagent-app-v2%3FVDIPlatform%3DCITRIX

Step 3 — Build or Deploy an Audio-Optimized CCP

If you are using a custom CCP (not the native Amazon Connect CCP), you need to use the Amazon Connect Streams.js and Connect-RTC.js libraries and pass the VDI platform parameter in your initialization code:initCCP.js — Key Configuration

connect.core.initCCP(
  document.getElementById(ccpContainerId), {
    ccpUrl: connectCCPUrl,           // Your CCP URL
    loginPopup: true,
    region: connectRegion,
    softphone: {
      allowFramedSoftphone: true,
      VDIPlatform: "CITRIX",         // Key parameter for Citrix
    },
    pageOptions: {
      enableAudioDeviceSettings: false,  // Disable for Embedded CCP
      enablePhoneTypeSettings: false,
    }
  }
);

AWS provides a sample project on GitHub that you can clone and deploy as a starting point: aws-samples/amazon-connect-audio-optimization. It includes both an Embedded CCP and a Custom CCP sample with all the necessary Citrix UC SDK integration already wired up.

Step 4 — Approve the CloudFront URL in Amazon Connect

If you host your custom CCP on Amazon CloudFront (which the AWS sample project does by default), you must add the CloudFront domain to your Amazon Connect instance’s approved origins list:

Go to the AWS Console → Amazon Connect → your instance.
In the left menu, click Approved origins.
Click Add domain and enter your CloudFront URL (e.g., https://xxxxxxxx.cloudfront.net).
Save. Without this step, your CCP will silently fail to load with no useful error message.

Step 5 — Update the Agent Security Profile

When using an Embedded CCP, you need to prevent agents from changing their audio devices through the CCP interface (since audio device selection is now managed by the Citrix UC SDK on the thin client):

In your Amazon Connect console, go to Users → Security Profiles → your Agent profile, and make sure Audio device settings is unchecked. This step is mandatory for Embedded CCP but optional for Custom CCP.

Check – KloudMint’s Amazon Connect Agent Workstation Validator Tool

Fix: Audio Issues on Amazon WorkSpaces

WorkSpaces audio troubleshooting follows a different path depending on whether you are using the PCoIP or DCV protocol. Here is how to handle each one.

Fixing Audio on PCoIP WorkSpaces

The most common issue with PCoIP WorkSpaces is a missing or inactive Teradici Virtual Audio Driver. Without this driver, Windows cannot properly route audio through the VDI session.

To check if the driver is present, open Device Manager → Sound, video and game controllers. If you do not see “Teradici Virtual Audio Driver” listed, install it:

Open Device Manager → Action → Add legacy hardware.
Choose Install hardware that I manually select from a list (Advanced).
Select Show All Devices, then click Have Disk and browse to: C:\Program Files\Teradici\PCoIP Agent\drivers\TeraAudio
Select Teradici Virtual Audio Driver, complete the wizard, then reboot.

After the reboot, verify the driver is set as the active playback device: right-click the speaker icon in the taskbar → Sounds → Playback tab → look for the green checkmark next to Teradici Virtual Audio Driver. Repeat the same check on the Recording tab for microphone input.

Fixing Audio on DCV WorkSpaces

For DCV protocol WorkSpaces, you need two separate drivers: one for speakers and one for the microphone.

Navigate to C:\Program Files\NICE\DCV\Server\drivers\audio and install:

AWSVirtualSpeakersExt.inf — for audio playback (AWS Virtual Speakers 7.1 Device)
AWSVirtualMicrophone.inf — for microphone input (AWS Virtual Microphone Device)

Install both using the same Device Manager → Add Legacy Hardware process described above for PCoIP, and confirm each appears with a green checkmark as the active playback and recording device respectively.

The Windows Microphone Privacy Setting Trap

Windows Server 2019 and later versions have a privacy setting that turns off microphone access by default. This catches a surprising number of people. Even if all your drivers are installed correctly, apps cannot detect a microphone if this setting is off.

To fix it: Settings → Privacy → Microphone → “Allow apps to access your microphone” → On.

Don’t skip this: This Windows privacy setting is turned off by default on WorkSpaces Windows Server images. It is easy to miss, and it will make your microphone completely invisible to the browser and CCP, even with all drivers installed and configured correctly.

Network Setup That Most People Get Wrong

Network configuration is the single biggest cause of Amazon Connect VDI audio quality problems. The good news is the requirements are clear. The bad news is there are two separate machines to configure, and most people only configure one of them.

The Two Machines You Need to Configure

Think of the architecture as two separate traffic lanes:

Machine	What it needs to reach	Protocol & Port	Direction
Citrix VDA Server	Amazon Connect signalling & CloudFront	TCP 443 (HTTPS + WebSocket)	Outbound
Agent Thin Client	Amazon Connect Softphone Media	UDP 3478	Outbound (send/receive)

Citrix VDA — Allow These Domains on TCP 443

URL Allowlist for Citrix VDA (TCP 443)

*.transport.connect.{region}.amazonaws.com
*.telemetry.connect.{region}.amazonaws.com
{yourInstanceName}.my.connect.aws
*.static.connect.aws
*.cloudfront.net

Replace {region} with your Amazon Connect instance region (e.g., us-east-1, ap-southeast-2, eu-west-2).

Agent Thin Client — Allow UDP 3478

URL Allowlist for Agent Machine (UDP 3478)

TurnNlb-*.elb.{region}.amazonaws.com

Port:      3478 (UDP)
Direction: Outbound (SEND and RECEIVE)

This is the #1 missed fix: UDP port 3478 must be open outbound from the agent’s thin client — not the VDA server. If your corporate firewall or VPN blocks outbound UDP, all audio traffic falls back to routing through the VDA, which is exactly the problem you are trying to solve. If agents are on VPN, you may need a split-tunnel configuration that exempts UDP 3478 traffic.

Checking Packet Loss — The ToInstancePacketLossRate Metric

If audio quality is degraded but connectivity seems fine, check your Amazon Connect packet loss metric in CloudWatch:

Open the CloudWatch console and go to Metrics → All metrics → Connect.
Find your Amazon Connect instance ID and select ToInstancePacketLossRate.
If the value exceeds 0.2 (20%), escalate to your network team — there is a local network issue dropping packets before they reach Amazon Connect.

Windows Exclusive Mode Interference

This is a lesser-known cause of microphone issues: Windows Exclusive Mode. When enabled, one application can take sole control of an audio device, blocking the CCP from accessing it. Check this in your audio device properties and ensure “Allow applications to take exclusive control of this device” is either disabled or not conflicting with the browser.

Also Check – How to Fix “Failed to Get Email Message” in Amazon Connect (S3 CORS Policy)

How to Verify Audio Redirection Is Working

After making your configuration changes, you need to confirm that audio is actually flowing from the thin client and not the VDA server. There are three ways to verify this.

Method 1 — Task Manager on the Thin Client

Open Task Manager on the agent’s thin client (not the VDA session) and look for the following processes in the background processes list:

HDX RTC Engine (32 bit) — this is the Citrix WebRTC redirection engine
Citrix HDX Engine (32 bit)

If you can see these processes running on the thin client during a call, audio redirection is active. If these processes are absent, audio is still being processed server-side.

Method 2 — The vdiClientConnected Event in CCP Logs

In your CCP logs (either download via the embedded CCP “Download logs” button, or use connect.getLog().download() in a custom CCP), look for the vdiClientConnected event. This event is dispatched when Citrix WebRTC redirection initialises successfully:CCP Log — Successful VDI Connection

{
  "event": "vdiClientConnected",
  "type_script": { "type": 5, "version": "1.6.0.1" },
  "webrpc":      { "type": 0, "version": "1.6.0." },
  "webrtc_codecs": { "type": 1, "version": "1.76.0.0" },
  "receiver":    { "type": 2, "version": "19.12.0.14" }
}

If you see vdiClientDisconnected instead, check the reason field for a clue:

Reason Code	Meaning	Fix
`"Websocket Connection Failure"`	WebSocket from VDA to thin client failed	Check CtxHdxWebSocketService is running; verify registry key is set
`"endpointDisconnected"`	HdxRtcEngine.exe crashed or Citrix session disconnected	Check thin client stability; update Citrix Workspace App
`"endpointUnsupported"`	Citrix Workspace App version too old	Upgrade CWA to 2305 or later
`"Citrix Webrtc Redir Reg Key not Present"`	Registry key missing or not applied correctly	Re-add the ProcessWhitelist key and restart CtxHdxWebSocketService

Method 3 — Wireshark Packet Capture on the Thin Client

For a definitive test, install Wireshark on the agent’s thin client and capture traffic on the local network interface. Apply the filter udp.port == 3478 and place a test call. If you see UDP traffic flowing on port 3478, the audio is being sent from the thin client directly to Amazon Connect. If you see no UDP 3478 traffic on the thin client during a call, the audio is still routing through the VDA.

Common Errors and What They Mean

One-Way Audio

The agent can hear the customer but the customer cannot hear the agent (or vice versa). This is almost always a firewall issue. Specifically, UDP 3478 is often blocked in one direction. Verify that both send and receive are allowed, not just outbound. Also check NAT configuration — if your network uses NAT, make sure keep-alives are configured so the connection does not drop mid-call.

Audio Cutting Out Every Few Seconds

Intermittent audio dropout is usually a packet loss or bandwidth issue. Check the ToInstancePacketLossRate metric in CloudWatch. Also look for VPN-related issues — encrypted VPN tunnels add latency and overhead that can cause periodic audio gaps. If agents on VPN have worse audio than those without, split-tunnel VPN is the fix.

Echo on Agent’s Side

Echo usually means the agent’s microphone is picking up audio from their speakers. The fix is almost always a headset — use a headset with acoustic echo cancellation (AEC). Citrix adaptive audio includes an echo cancellation feature that is enabled by default starting with CWA version 2411. Also check that speakers and microphone are not too close to each other physically.

Audio Delays of 2 Seconds or More

If you are seeing multi-second delays, audio redirection is almost certainly not working — audio is still going through the VDA. Work through the verification steps above to confirm redirection status and check the vdiClientConnected event in your CCP logs.

Calls Work but Audio Quality Degrades Over Time

This is commonly caused by high server CPU load. When multiple agents handle concurrent calls and audio is being processed on the VDA (because redirection is not set up), the server becomes resource-constrained and audio quality degrades under load. Enable audio redirection to move processing off the server, or increase your VDA session density configuration.

Citrix Audio Quality Policy Settings Explained

Even with audio redirection working, you may want to tune the Citrix audio quality policy settings to get the best balance of audio quality and bandwidth for your environment.

Adaptive Audio (Default — Recommended)

Adaptive audio is the default and the best starting point for most deployments. It automatically adjusts audio settings based on network conditions, so you do not need to manually tune codecs for different network types. It uses variable bitrate in stereo for playback and mono for microphone capture at 48000 Hz.

Recommendation: For real-time voice communications (like Amazon Connect), always use the Medium quality (Optimized for speech) policy setting regardless of your transport protocol. The High Definition audio setting is not suitable for softphone use — it adds latency that disrupts real-time conversation.

Also Check – Fixing ‘No Audio’ in Amazon Connect: The Ultimate Amazon Connect WebRTC error 503 Citrix fix (2026 Update)

Bandwidth Guidelines at a Glance

Quality Level	Bitrate (Approx)	Codec	Best For
Adaptive (default)	Variable	Automatic	Most deployments; adjusts to network
High (HD Audio)	75–175 kbps	44100 Hz stereo	Music, not VoIP — introduces latency
Medium (VoIP speech)	16–40 kbps	16000 Hz mono	✅ Best for Amazon Connect agent calls
Low	10–25 kbps	8000 Hz mono	Dial-up / very constrained bandwidth

UDP vs TCP for Audio Transport

Citrix strongly recommends using UDP transport for real-time audio. Unlike TCP, UDP does not retransmit dropped packets — which sounds like a problem, but for voice it is actually better. A retransmitted packet that arrives 200ms late causes an audible gap; a dropped packet simply results in a brief quiet moment that the human ear compensates for automatically. TCP retransmissions under packet loss conditions introduce noticeable, disruptive latency spikes.

Enable Audio over UDP Real-time Transport in your Citrix policy. The default port range is 16500–16509 on the VDA side. Make sure these ports are not blocked by your internal firewall.

Full Amazon Connect VDI Audio Quality Fix Checklist

Use this checklist systematically when diagnosing audio quality problems. Work through it in order — each item eliminates a category of problem.

Carrier and Amazon Connect Side

Test with multiple carriers.Call your Amazon Connect number from different telecom carriers. If only one carrier has issues, the problem is carrier-side, not your configuration.
Check call recordings.Amazon Connect stores customer audio on the left channel and agent audio on the right channel. If only one channel has issues, that tells you where the problem originates.
Include contact IDs when opening support cases.Always collect 3–4 affected contact IDs from within 48 hours of the issue, along with timestamps and timezone.

Agent Hardware and Software

Test with a different headset.If audio quality improves with a different headset, the original headset is the issue.
Test with a different workstation.Isolates whether the issue is machine-specific.
Confirm browser microphone permissions.Use the Amazon Connect Endpoint Test Utility to check microphone access in the browser.
Disable Windows exclusive mode.Check audio device properties and ensure no application has exclusive control of the microphone.
Enable microphone privacy (WorkSpaces).Settings → Privacy → Microphone → On.

VDI and Audio Redirection

Citrix registry key present on VDA.ProcessWhitelist under WebSocketService with Chrome.exe and msedge.exe.
CtxHdxWebSocketService running on VDA.Check in Task Manager → Services tab.
CWA version 2305 or later on thin client.Check Citrix Workspace App About screen.
VDA version 2203 LTSR or later.Older versions do not support WebRTC redirection.
VDIPlatform query parameter in agent workspace URL.Must match the VDI environment (CITRIX or AWS_WORKSPACE).
HDX RTC Engine process visible in thin client Task Manager during calls.Confirms redirection is active.
vdiClientConnected event in CCP logs.No vdiClientDisconnected events.

Network and Firewall

UDP 3478 outbound open from thin client.Both send and receive to TurnNlb-*.elb.{region}.amazonaws.com.
TCP 443 open from VDA to Amazon Connect domains.Including *.transport.connect, *.telemetry.connect, and CloudFront.
VPN split-tunnelling configured.If agents use VPN, UDP 3478 traffic must bypass VPN encryption overhead.
ToInstancePacketLossRate below 0.2 in CloudWatch.If above, escalate to network team.
CloudFront domain in Amazon Connect Approved Origins.If using a hosted Custom CCP.

Also Check – The Ultimate Guide to Amazon Connect AI Agent Content Segmentation

FAQs about Amazon Connect VDI audio quality fix

Can I use the native Amazon Connect CCP with Citrix audio optimization?

No. Audio optimization is only supported with an Embedded or Custom CCP built using the Amazon Connect Streams.js and Connect-RTC.js libraries. The native CCP (accessed directly at your-instance.my.connect.aws/ccp-v2) does not support VDI audio redirection. Media will continue to flow through the browser inside the VDA server, which is the source of the problem. You need to build or deploy a custom CCP.

My agents are on VPN. Does that block UDP 3478?

It depends on how your VPN is configured. Many corporate VPN setups route all traffic through the VPN tunnel, which adds latency and can interfere with UDP. The best solution is to configure split-tunnel VPN so that traffic to Amazon Connect endpoints (TurnNlb-*.elb.{region}.amazonaws.com on UDP 3478) goes directly to the internet rather than through the VPN tunnel. This significantly reduces audio latency for remote agents.

How do I check if audio redirection is working without Wireshark?

The easiest method is Task Manager on the thin client. During a live call, open Task Manager and look for “HDX RTC Engine (32 bit)” in the background processes. If it is running, audio redirection is active. Alternatively, download the CCP logs and search for the “vdiClientConnected” event — its presence confirms successful WebRTC redirection.

Does Amazon Connect audio optimization work with Salesforce CTI Adapter?

No. The Salesforce CTI Adapter does not support VDI platform detection, which means audio cannot be redirected through the UC SDK. When agents use the Salesforce CTI Adapter in a VDI environment, media continues to route through the VDA browser, resulting in the same audio quality issues. To get audio optimization, you need to use the Amazon Connect agent workspace or a custom CCP instead of the Salesforce CTI Adapter.

Is there a difference between Citrix VDI audio fix and Amazon WorkSpaces audio fix?

Yes, the approach is different. For Citrix, you use the Citrix UC SDK via the Connect-RTC.js library, which communicates over the ICA virtual channel to redirect WebRTC audio to the thin client. For Amazon WorkSpaces, audio optimization is built into the WorkSpaces client protocol (PCoIP or DCV), and you need to ensure the correct virtual audio drivers are installed and configured as the active playback and recording devices. Both end up achieving the same goal — audio processed on the local device rather than the remote desktop server.

What should I include when opening an AWS Support case for audio issues?

Include: source phone number, destination phone number, 3–4 affected contact IDs (within 48 hours of the issue), timestamps with timezone, a description of the specific audio issue (delay, one-way, choppy, echo), call recordings if available, and your VDI configuration details (VDA version, CWA version, CCP type). The more context you provide, the faster AWS Support can diagnose the root cause.

Can agents switch audio devices during a live call in Citrix?

This is not recommended and is something to avoid. Because media device information is set at the start of a contact, switching audio devices mid-call means the agent cannot access audio from the new device for the current call. Agents should stay with one audio device for the entire duration of a contact. If agents need to use a specific device, ensure it is selected before the call begins.

Final Thoughts on Amazon Connect VDI Audio Quality Fix

The root cause of almost every Amazon Connect VDI audio quality issue comes down to one thing: audio is routing through the VDA server when it should be going directly from the thin client to Amazon Connect. Every fix in this guide ultimately serves that one goal.

If you only do three things from this entire guide, do these: open UDP port 3478 on the thin client firewall, add the Citrix WebSocket registry key on the VDA, and append the VDIPlatform query parameter to your agent workspace URL. Those three changes resolve the majority of Amazon Connect VDI audio lag issues.

For anything more complex — persistent packet loss, VPN interference, or Salesforce CTI limitations — the troubleshooting checklist and CloudWatch metrics will help you pinpoint the specific layer where the problem lives. If you are still stuck after working through this guide, open an AWS Support case with the contact IDs and let the team run a proper network trace.

Sources – AWS Doc for Citrix, AWS Doc for Workspace