Hi everyone,
Looking for advice / experiences from anyone who has run 3CX in a similar setup.
**Setup:**
- Windows Server 2025 Datacenter host with Hyper-V role
- One VM per user (Windows desktop OS), reached via RDP from a thin client
- 3CX Windows softphone installed inside the VM
- USB headset plugged into the thin client, redirected to the VM via RDP audio
- vSwitch External; VM and thin client on the same LAN
- SIP trunk to an external provider
- RTP capture on the host shows healthy stream (low jitter, no packet loss to the provider)
**Symptoms (from the remote party's perspective, NOT the local user):**
- **Noticeable latency** in the conversation — people talk over each other, clearly above the 150 ms one-way budget
- **Background noise** on the audio the remote party receives (the local user hears the remote party cleanly)
So the problem is specifically on the **outbound** path: mic → thin client → RDP audio capture redirection → VM → 3CX → RTP to provider.
**What I already ruled out:**
- Headset hardware: same headset on a regular PC works perfectly, no noise
- Network to provider: RTP capture looks clean, jitter < 30 ms, no loss
- The local user doesn't hear noise on the incoming audio, so the inbound RTP and RDP playback redirection seem fine
**My current understanding:**
The outbound audio chain is accumulating both delay and noise because RDP audio capture redirection isn't designed for real-time voice. The microphone audio gets buffered by RDP, then handed to the 3CX client, then through Hyper-V vSwitch, then out as RTP. Each stage adds latency, and the redirection seems to be degrading the captured audio.
**Questions:**
1. Is anyone successfully running the 3CX Windows client inside an RDP session in production? If yes, what optimizations did you apply (RDP audio quality GPO, codec choice, NS/AGC in 3CX vs Windows, Hyper-V VMQ off, etc.)?
2. Is the official recommendation from 3CX to avoid softphones inside RDP sessions altogether? I've seen comments suggesting this but no clear KB article.
3. For thin client environments, what's the recommended deployment?
- 3CX Web Client on the thin client itself (not inside RDP)
- Native SIP client on the thin client
- Physical SIP desk phone
4. Has anyone moved from "3CX inside RDP" to one of the above and confirmed the latency dropped?
Happy to share captures or specific 3CX settings if it helps. Thanks!
Looking for advice / experiences from anyone who has run 3CX in a similar setup.
**Setup:**
- Windows Server 2025 Datacenter host with Hyper-V role
- One VM per user (Windows desktop OS), reached via RDP from a thin client
- 3CX Windows softphone installed inside the VM
- USB headset plugged into the thin client, redirected to the VM via RDP audio
- vSwitch External; VM and thin client on the same LAN
- SIP trunk to an external provider
- RTP capture on the host shows healthy stream (low jitter, no packet loss to the provider)
**Symptoms (from the remote party's perspective, NOT the local user):**
- **Noticeable latency** in the conversation — people talk over each other, clearly above the 150 ms one-way budget
- **Background noise** on the audio the remote party receives (the local user hears the remote party cleanly)
So the problem is specifically on the **outbound** path: mic → thin client → RDP audio capture redirection → VM → 3CX → RTP to provider.
**What I already ruled out:**
- Headset hardware: same headset on a regular PC works perfectly, no noise
- Network to provider: RTP capture looks clean, jitter < 30 ms, no loss
- The local user doesn't hear noise on the incoming audio, so the inbound RTP and RDP playback redirection seem fine
**My current understanding:**
The outbound audio chain is accumulating both delay and noise because RDP audio capture redirection isn't designed for real-time voice. The microphone audio gets buffered by RDP, then handed to the 3CX client, then through Hyper-V vSwitch, then out as RTP. Each stage adds latency, and the redirection seems to be degrading the captured audio.
**Questions:**
1. Is anyone successfully running the 3CX Windows client inside an RDP session in production? If yes, what optimizations did you apply (RDP audio quality GPO, codec choice, NS/AGC in 3CX vs Windows, Hyper-V VMQ off, etc.)?
2. Is the official recommendation from 3CX to avoid softphones inside RDP sessions altogether? I've seen comments suggesting this but no clear KB article.
3. For thin client environments, what's the recommended deployment?
- 3CX Web Client on the thin client itself (not inside RDP)
- Native SIP client on the thin client
- Physical SIP desk phone
4. Has anyone moved from "3CX inside RDP" to one of the above and confirmed the latency dropped?
Happy to share captures or specific 3CX settings if it helps. Thanks!
