network foundations · published

Troubleshooting audio delay and talk-over on voice calls

Verified 2026-10-02 · 56 sources · tier 1–2

Also known as audio-delay, talker-overlap.

Troubleshooting audio delay and talk-over on voice calls

Amazon Connect documentation lists delayed audio as a call-quality symptom whose effect is consistent overlapping in conversation between caller and agent, and attributes it to constrained bandwidth or to hardware or workstation congestion 2. ITU-T Recommendation G.114, titled One-way transmission time, is the May 2003 edition and is listed by ITU as in force 31.

Delay limits and talk-over

Cisco's packet-voice delay technote restates G.114 as classifying 0 to 150 ms one-way delay as acceptable for most user applications 14. The same technote states that 150 to 400 ms one-way delay is acceptable provided administrators are aware of the transmission time and its impact on quality 15. Cisco restates G.114 as treating one-way delay above 400 ms as unacceptable for general network planning purposes, while recognising exceptional cases 17.

Cisco notes that the G.114 delay limits assume echo is adequately controlled, meaning echo cancellers are in use 16. Citing G.131, Cisco's technote states that echo cancellers are required when one-way delay exceeds 25 ms, so added latency makes residual echo more noticeable 12.

Cisco cites satellite connections as a source of talker overlap, giving satellite delay as roughly 500 ms (250 ms up and 250 ms down) 20.

Where the delay comes from

Cisco classifies coder (processing), packetization, serialization and the fixed part of network switching delay as fixed delay components in a voice delay budget 13. Cisco classifies queuing and buffering delay as variable delay, which the receiving end absorbs with a de-jitter buffer 23. Cisco's worked single-hop delay budget example totals about 138 ms one-way 21.

Cisco's technote lists G.729A coder delay as 2.5 ms best case and 10 ms worst case, compared with 5 ms and 20 ms for G.723.1 18.

The de-jitter buffer

Cisco describes the de-jitter buffer as transforming variable network delay into a fixed delay by holding the first packet for an initial play-out delay 11. According to Cisco, the buffer's contribution to end-to-end delay is its initial play-out delay plus however long the first packet was itself buffered in the network 10. Microsoft's archived media-quality guidance states that buffering for jitter increases end-to-end latency 47. Azure Communication Services adapts to some jitter through buffering, and participants notice jitter only when it exceeds the buffering 1.

On Cisco IOS voice gateways, the de-jitter (playout delay) buffer runs in adaptive mode by default, and the playout-delay command can change its dynamic behaviour or set fixed mode 19.

Media hops that decode and re-encode

In Cisco's two-hop tandem example, the voice is decompressed and de-jittered, then recompressed and de-jittered a second time 22. WarmTransfer's reading of the sources is that each media hop that decodes and re-encodes audio, such as a transcoder or a B2BUA media processor, adds another coder, packetization and de-jitter stage to the one-way delay, so counting such hops is a first step when diagnosing talk-over 50.

In Teams Direct Routing, the Media Processor is a B2BUA that can transcode (for example SILK to G.711), whereas Transport Relays only relay audio and never change the codec 29. Teams Direct Routing media bypass keeps media between the SBC and the Teams client to shorten the media path and reduce hops 28. Teams Media Processors are always in the media path for voice applications such as Call Park, Auto Attendant and Call Queues, and for group escalation, federated calls and Teams web clients, even when the trunk has media bypass enabled 30.

Measuring delay with RTCP

RTCP lets an RTP sender estimate round-trip propagation delay from a receiver report as arrival time minus LSR minus DLSR, where DLSR is expressed in units of 1/65536 seconds 39. RFC 3550 notes that this estimate is the round-trip delay at the time the sender composed the SR, and that later queuing and processing could make the current delay longer 40. RFC 3550 defines interarrival jitter as the mean deviation (smoothed absolute value) of the difference in packet spacing at the receiver compared with the sender, measured in RTP timestamp units 38.

The RTCP XR VoIP Metrics report block carries a round trip delay field, which holds the most recently calculated round trip time between RTP interfaces in milliseconds 44. The same block reports end system delay separately from network round trip delay, in milliseconds 42. It also reports the current nominal jitter buffer delay and the current maximum jitter buffer delay, both in milliseconds 43. The RTCP XR DLRR report block can be sent in response to a Receiver Reference Time block so that a receiver that is not sending media can calculate its round trip time 41.

For its Azure-based network assessment tables, the archived Microsoft guidance states that one-way latency targets are half of the corresponding RTT targets 48. We infer that halving an RTCP round-trip figure gives only the network part of one-way delay and assumes a symmetric path. Because mouth-to-ear delay also includes end-system delay such as codec and jitter buffer, it will be higher than RTT/2 45.

Platform thresholds

Each threshold below is the stated position of the named source. The thresholds are not interchangeable, because they measure different segments of the call.

Source and metric	Stated threshold	Notes
Teams CQD, average Round Trip	Stream marked Poor above 500 ms, for streams with more than 500 packets 25	Computed as described in RFC 3550, in ms 27
Teams CQD, average jitter and packet loss rate	Poor when jitter exceeds 30 ms or packet loss rate exceeds 0.1 24
Archived Skype for Business Online, client to Microsoft network edge	Round trip under 60 ms optimal and over 500 ms poor; jitter under 3 ms optimal and over 30 ms poor 46	Assumed sites and edges on the same continent 49
Amazon Connect Endpoint Test Utility, instance to agent browser latency	Average should not exceed 300 ms 3
Amazon Connect QualityMetrics, jitter buffer	Above 30 ms, or changing very frequently, indicates network congestion or low router bandwidth 5	Flagged as HighJitterBuffer 5
Webex Calling Control Hub, end-to-end	Good when packet loss is under 5%, latency or RTT is under 400 ms and jitter is under 150 ms 55
Webex Calling Control Hub, single media relay hop	Good when packet loss is under 2.5%, latency or RTT is under 200 ms and jitter is under 75 ms 56
Genesys WebRTC Diagnostics, round-trip average	Below 300 ms, equivalent to 150 ms one-way 36
Genesys WebRTC Diagnostics, jitter and packet loss	Jitter under 30 ms and packet loss under 1 percent 35
Genesys Cloud, phone to Edge one-way latency	Under 150 ms on WAN or Internet, under 75 ms on the same LAN as the Edge 33
Genesys Cloud, Edge to Genesys Cloud	Under 300 ms round trip and under 5% packet loss 32

Microsoft states that a Poor audio classification in CQD does not mean that the stream was actually of poor quality or that the user perceived a problem 26.

Diagnosing it

Check where the call was anchored

AWS states that an agent physically distant from the AWS Region hosting the Amazon Connect instance will see increased RoundTripTime 7. WarmTransfer's reading of the sources is that, across cloud platforms, a media region or edge far from the user adds round-trip time directly. When callers talk over each other, check which media region or edge anchored the call before tuning QoS 37.

Twilio's roaming edge uses Global Low Latency routing to select the data center with the lowest-latency connection to the user, and Voice/SIP customers can instead pin a named edge location 54. Genesys recommends testing Global Media Fabric configurations to balance audio latency against call setup time, and notes that results vary with carrier location, agent location and internet quality 34.

Check VPN and virtual desktop paths

Microsoft recommends a split-tunnel VPN so that Teams traffic bypasses the VPN and goes directly to Microsoft 365 52. Microsoft warns that VPN routing can send Teams traffic to a service front door farther from the user, which introduces extra latency and jitter, and can hairpin traffic through the VPN device 53. Amazon Connect network troubleshooting guidance states that encrypted VPN tunnels add overhead and latency that degrade audio 9. AWS states that routing Amazon Connect audio through a virtual desktop, instead of redirecting the WebRTC session to the agent workstation, can introduce high round trip time 8.

Pull per-call metrics

Amazon Connect contact records can flag HighRoundTripTime in QualityMetrics PotentialQualityIssues, and AWS states that high RTT makes callers experience noticeable delays (speech overlap) 4. QualityMetrics is returned by the DescribeContact API and in the Kinesis contact record stream. It is not shown when viewing the contact record in the admin website and is not part of EventBridge events 6.

For Webex Calling, compare the end-to-end grade with the per-hop grade, because Control Hub applies different thresholds to each: under 400 ms latency or RTT end-to-end and under 200 ms per media relay hop 5556.

Count transcoding hops

In Teams Direct Routing, confirm whether the call type keeps the Media Processor in the path despite media bypass, as it does for Call Park, Auto Attendant, Call Queues, group escalation, federated calls and Teams web clients 30. We infer that each decode-and-re-encode hop found this way adds coder, packetization and de-jitter delay to the one-way total 50.

Applicability

Applies to: AWS Amazon Connect, Cisco, Microsoft Skype for Business Online, Microsoft Azure Communication Services, Cisco IOS voice gateways, Microsoft Teams Direct Routing, Microsoft Teams, Cisco Webex Calling, Genesys Cloud, and Twilio Programmable Voice. Deployments: on-premises, hybrid, multi-tenant, and any. Sources checked 2026-10-02. The Cisco delay technote cited throughout is dated 2006-02-02 14. The Cisco IOS playout-delay behaviour applies to on-premises Cisco IOS voice gateways 19. The Skype for Business Online targets come from archived 2017 Microsoft guidance for a retired product 46. The Genesys Cloud phone-to-Edge and Edge connectivity requirements apply to hybrid deployments 3332. The Teams, Amazon Connect, Webex Calling, Twilio, Azure Communication Services and Genesys Cloud WebRTC material reflects multi-tenant services as of 2026-10-02 5245554136.

What remains uncertain

As of 2026-10-02, the current Microsoft Teams Prepare your organization's network page publishes bandwidth figures but no latency, round-trip or jitter target table 51. The Teams round-trip figure that remains documented is the CQD Poor classification above 500 ms 25.

How Webex Calling selects media regions and relays is not covered by the sources below. The contents of Webex's separate network requirements article are also not covered by the sources below.

Call quality thresholds for cloud calling platforms other than those named in this article are not covered by the sources below.

The clause text of the G.114 PDF itself is not covered by the sources below. The delay bands given here are Cisco's restatement of G.114 14.

The measured latency that a specific call-control platform adds when it inserts a transcoder or media termination point is not covered by the sources below.

Voice latency over low-earth-orbit satellite links is not covered by the sources below. The only satellite figure here is Cisco's roughly 500 ms for satellite connections, which the claim's scope places in a geostationary context 20.

The mechanism by which Genesys Global Media Fabric selects media regions and TURN servers is not covered by the sources below.

Sources

1
Azure Communication Services adapts to some jitter through buffering, and participants notice jitter only when it exceeds the buffering.
Network recommendations - Azure Communication Services · Network quality, Inter-packet arrival jitter bullet · Checked 2026-10-02
2
Amazon Connect documentation lists delayed audio as a call-quality symptom whose effect is consistent overlapping in conversation between caller and agent, attributing it to constrained bandwidth or hardware or workstation congestion.
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Symptoms of call quality issues > Delayed audio · Checked 2026-10-02
3
Amazon Connect's Endpoint Test Utility measures latency between the instance and the agent browser, and AWS says average latency should not exceed 300 ms.
Troubleshoot your network for call quality and disconnect problems · Run the Endpoint Test Utility, first bullet · Checked 2026-10-02
4
Amazon Connect contact records can flag HighRoundTripTime in QualityMetrics PotentialQualityIssues, and AWS states high RTT makes callers experience noticeable delays (speech overlap).
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Use QualityMetrics > HighRoundTripTime · Checked 2026-10-02
5
Amazon Connect guidance says a jitter buffer above 30 ms, or one changing very frequently, indicates network congestion or low router bandwidth (HighJitterBuffer).
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Use QualityMetrics > HighJitterBuffer > Causes · Checked 2026-10-02
6
Amazon Connect QualityMetrics is returned by the DescribeContact API and in the Kinesis contact record stream, but is not shown when viewing the contact record in the admin website and is not part of EventBridge events.
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Where to find QualityMetrics · Checked 2026-10-02
7
AWS states that an agent physically distant from the AWS Region hosting the Amazon Connect instance will see increased RoundTripTime.
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Use QualityMetrics > HighRoundTripTime > Causes · Checked 2026-10-02
8
AWS states that routing Amazon Connect audio through a virtual desktop, instead of redirecting the WebRTC session to the agent workstation, can introduce high round trip time.
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Use QualityMetrics > HighRoundTripTime > Causes, last bullet · Checked 2026-10-02
9
Amazon Connect network troubleshooting guidance states that encrypted VPN tunnels add overhead and latency that degrade audio.
Troubleshoot your network for call quality and disconnect problems · Investigate network components and devices, VPNs bullet · Checked 2026-10-02
10
Per Cisco, the de-jitter buffer's contribution to end-to-end delay is its initial play-out delay plus however long the first packet was itself buffered in the network.
Understanding Delay in Packet Voice Networks · De-Jitter Delay · Checked 2026-10-02
11
Cisco describes the de-jitter buffer as transforming variable network delay into a fixed delay by holding the first packet for an initial play-out delay.
Understanding Delay in Packet Voice Networks · De-Jitter Delay · Checked 2026-10-02
12
Cisco's technote states, citing G.131, that echo cancellers are required when one-way delay exceeds 25 ms, so added latency makes residual echo more noticeable.
Understanding Delay in Packet Voice Networks · Standards for Delay Limits · Checked 2026-10-02
13
Cisco classifies coder (processing), packetization, serialization and the fixed part of network switching delay as fixed delay components in a voice delay budget.
Understanding Delay in Packet Voice Networks · Sources of Delay · Checked 2026-10-02
14
Cisco's packet-voice delay technote restates G.114 as classifying 0 to 150 ms one-way delay as acceptable for most user applications.
Understanding Delay in Packet Voice Networks · Standards for Delay Limits · Checked 2026-10-02
15
The same Cisco technote states that 150 to 400 ms one-way delay is acceptable provided administrators are aware of the transmission time and its impact on quality.
Understanding Delay in Packet Voice Networks · Standards for Delay Limits · Checked 2026-10-02
16
Cisco notes that the G.114 delay limits assume echo is adequately controlled, meaning echo cancellers are in use.
Understanding Delay in Packet Voice Networks · Standards for Delay Limits · Checked 2026-10-02
17
Cisco's technote restates G.114 as treating one-way delay above 400 ms as unacceptable for general network planning purposes, while recognising exceptional cases.
Understanding Delay in Packet Voice Networks · Standards for Delay Limits · Checked 2026-10-02
18
Cisco's technote lists G.729A coder delay as 2.5 ms best case and 10 ms worst case, versus 5 ms and 20 ms for G.723.1.
Understanding Delay in Packet Voice Networks · Coder (Processing) Delay table · Checked 2026-10-02
19
On Cisco IOS voice gateways the de-jitter (playout delay) buffer runs in adaptive mode by default, and the playout-delay command can change its dynamic behaviour or set fixed mode.
Understanding Jitter in Packet Voice Networks (Cisco IOS Platforms) · Introduction / playout delay buffer discussion · Checked 2026-10-02
20
Cisco cites satellite connections as a source of talker overlap, giving satellite delay as roughly 500 ms (250 ms up and 250 ms down).
Understanding Delay in Packet Voice Networks · Standards for Delay Limits (talker overlap paragraph) · Checked 2026-10-02
21
Cisco's worked single-hop delay budget example totals about 138 ms one-way.
Understanding Delay in Packet Voice Networks · Build the Delay Budget > Single-Hop Connection · Checked 2026-10-02
22
In Cisco's two-hop tandem example the voice must be decompressed and de-jittered, then recompressed and de-jittered a second time.
Understanding Delay in Packet Voice Networks · Two Hops on a Public Network with a C7200 that Acts as a Tandem Switch · Checked 2026-10-02
23
Cisco classifies queuing and buffering delay as variable delay, which the receiving end absorbs with a de-jitter buffer.
Understanding Delay in Packet Voice Networks · Queuing/Buffering Delay; De-Jitter Delay · Checked 2026-10-02
24
CQD marks an audio stream Poor when average jitter exceeds 30 ms, and also when packet loss rate exceeds 0.1.
Stream classification in Call Quality Dashboard (CQD) · Audio Classifier table, Jitter and Packet Loss Rate rows · Checked 2026-10-02
25
Teams Call Quality Dashboard marks an audio stream Poor if its average Round Trip exceeds 500 ms (or another audio threshold is broken) and the stream has more than 500 packets.
Stream classification in Call Quality Dashboard (CQD) · Classifier Definitions > Audio Classifier table, Round Trip row · Checked 2026-10-02
26
Microsoft states that a Poor audio classification in CQD does not mean the stream was actually of poor quality or that the user perceived a problem.
Stream classification in Call Quality Dashboard (CQD) · Audio Classifier, Note below the table · Checked 2026-10-02
27
CQD's Round Trip metric is the average round-trip network propagation time in milliseconds, computed as described in RFC 3550.
Stream classification in Call Quality Dashboard (CQD) · Audio Classifier table, Round Trip row, Explanation column · Checked 2026-10-02
28
Teams Direct Routing media bypass keeps media between the SBC and the Teams client to shorten the media path and reduce hops.
Plan for media bypass with Direct Routing · About media bypass with Direct Routing, first paragraph · Checked 2026-10-02
29
In Teams Direct Routing the Media Processor is a B2BUA that can transcode (for example SILK to G.711), whereas Transport Relays only relay audio and never change the codec.
Plan for media bypass with Direct Routing · Use of Media Processors and Transport Relays, comparison table and transcoding explanation · Checked 2026-10-02
30
Teams Media Processors are always in the media path for voice applications such as Call Park, Auto Attendant and Call Queues, and for group escalation, federated calls and Teams web clients, even when the trunk has media bypass enabled.
Plan for media bypass with Direct Routing · Use of Media Processors and Transport Relays; Use of Teams Media Processors if trunk is configured for media bypass; Requirements for using media processors · Checked 2026-10-02
31
ITU-T Recommendation G.114, titled One-way transmission time, is the May 2003 edition and is listed by ITU as in force.
G.114 : One-way transmission time · Recommendation landing page: Approved in / Status fields · Checked 2026-10-02
32
Genesys Cloud network readiness requires Edge connectivity to Genesys Cloud of under 300 ms round-trip latency and under 5% packet loss.
Customer network readiness (Genesys Cloud network requirements) · Network performance requirements, Edge connectivity row · Checked 2026-10-02
33
Genesys Cloud network readiness requires under 150 ms one-way latency between phone and Edge when the phone is on WAN or Internet, and under 75 ms when on the same LAN as the Edge.
Customer network readiness (Genesys Cloud network requirements) · Network performance requirements, external and internal phone call rows · Checked 2026-10-02
34
Genesys recommends testing Global Media Fabric configurations to balance audio latency against call setup time, noting results vary with carrier location, agent location and internet quality.
Global Media Fabric overview · Global Media Fabric overview, testing recommendation · Checked 2026-10-02
35
Genesys says the WebRTC Diagnostics jitter value should be under 30 ms and packet loss under 1 percent.
Run the standalone Genesys Cloud WebRTC Diagnostics app · Network Test results · Checked 2026-10-02
36
Genesys says the WebRTC Diagnostics round-trip average should be below 300 ms, equivalent to 150 ms one-way latency.
Run the standalone Genesys Cloud WebRTC Diagnostics app · Network Test results · Checked 2026-10-02
37
Across cloud platforms, a media region or edge far from the user adds round-trip time directly, so when callers talk over each other, check which media region or edge anchored the call before tuning QoS.inferred
Troubleshoot audio quality issues by using QualityMetrics in the contact record · Use QualityMetrics > HighRoundTripTime > Causes (region distance) · Checked 2026-10-02
38
RFC 3550 defines interarrival jitter as the mean deviation (smoothed absolute value) of the difference in packet spacing at the receiver compared with the sender, measured in RTP timestamp units.
RFC 3550 RTP: A Transport Protocol for Real-Time Applications · Section 6.4.1, interarrival jitter field; Appendix A.8 · Checked 2026-10-02
39
RTCP lets an RTP sender estimate round-trip propagation delay from a receiver report as arrival time minus LSR minus DLSR, where DLSR is expressed in units of 1/65536 seconds.
RFC 3550 RTP: A Transport Protocol for Real-Time Applications · Section 6.4.1, LSR and DLSR field definitions · Checked 2026-10-02
40
RFC 3550 notes the LSR/DLSR estimate is the round-trip delay at the time the sender composed the SR, and later queuing and processing could make the current delay longer.
RFC 3550 RTP: A Transport Protocol for Real-Time Applications · Section 6.4.1, note under the DLSR field · Checked 2026-10-02
41
The RTCP XR DLRR report block can be sent in response to a Receiver Reference Time block so that a receiver which is not sending media can calculate its round trip time.
RTP Control Protocol Extended Reports (RTCP XR) · Section 4.5, DLRR Report Block · Checked 2026-10-02
42
The RTCP XR VoIP Metrics block reports end system delay separately from network round trip delay, in milliseconds.
RTP Control Protocol Extended Reports (RTCP XR) · Section 4.7, end system delay · Checked 2026-10-02
43
The RTCP XR VoIP Metrics block reports the current nominal jitter buffer delay and the current maximum jitter buffer delay, both in milliseconds.
RTP Control Protocol Extended Reports (RTCP XR) · Section 4.7, JB nominal and JB maximum fields · Checked 2026-10-02
44
The RTCP XR VoIP Metrics report block carries a round trip delay field: the most recently calculated round trip time between RTP interfaces, in milliseconds.
RTP Control Protocol Extended Reports (RTCP XR) · Section 4.7, VoIP Metrics Report Block, round trip delay · Checked 2026-10-02
45
Halving an RTCP round-trip figure gives only the network part of one-way delay and assumes a symmetric path; mouth-to-ear delay also includes end-system delay such as codec and jitter buffer, so it will be higher than RTT/2.inferred
RTP Control Protocol Extended Reports (RTCP XR) · Section 4.7, round trip delay versus end system delay fields · Checked 2026-10-02
46
Archived Microsoft guidance (Skype for Business Online, 2017) set client-to-Microsoft-network-edge targets of round trip under 60 ms optimal and over 500 ms poor, with jitter under 3 ms optimal and over 30 ms poor.
Media Quality and Network Connectivity Performance - Skype for Business Online (retired) · Network Performance requirements from a Skype for Business client to Microsoft network Edge, metrics table · Checked 2026-10-02
47
Microsoft's archived media-quality guidance states that buffering for jitter increases end-to-end latency.
Media Quality and Network Connectivity Performance - Skype for Business Online (retired) · Factors that impact media quality > Network, Note · Checked 2026-10-02
48
For its Azure-based network assessment tables, the archived Microsoft guidance states that one-way latency targets are half of the corresponding RTT targets.
Media Quality and Network Connectivity Performance - Skype for Business Online (retired) · Measuring Network Performance using Azure VMs · Checked 2026-10-02
49
The archived Microsoft latency target assumed the customer sites and the Microsoft network edges are on the same continent.
Media Quality and Network Connectivity Performance - Skype for Business Online (retired) · Other performance target requirements, first bullet · Checked 2026-10-02
50
Each media hop that decodes and re-encodes audio (a transcoder or a B2BUA media processor) adds another coder, packetization and de-jitter stage to the one-way delay, so counting such hops is a first step when diagnosing talk-over.inferred
Understanding Delay in Packet Voice Networks · Effects of Multiple Compression Cycles; Two-hop delay budget examples · Checked 2026-10-02
51
As of 2026-10-02 the current Microsoft Teams Prepare your organization's network page publishes bandwidth figures but no latency, round-trip or jitter target table.
Prepare your organization's network for Microsoft Teams · Whole page: Network requirements, Network optimization, Bandwidth requirements · Checked 2026-10-02
52
Microsoft recommends a split-tunnel VPN so that Teams traffic bypasses the VPN and goes directly to Microsoft 365.
Prepare your organization's network for Microsoft Teams · Network optimization table, row Configure split-tunnel VPN · Checked 2026-10-02
53
Microsoft warns that VPN routing can send Teams traffic to a service front door farther from the user, introducing extra latency and jitter, and can hairpin traffic through the VPN device.
Prepare your organization's network for Microsoft Teams · Network optimization table, row Configure split-tunnel VPN, reasons list · Checked 2026-10-02
54
Twilio's roaming edge uses Global Low Latency routing to select the data center with the lowest-latency connection to the user, and Voice/SIP customers can instead pin a named edge location.
Edge Locations · Edge Locations page, edge list and roaming entry · Checked 2026-10-02
55
Webex Calling troubleshooting in Control Hub grades the end-to-end call experience as good when packet loss is under 5%, latency or RTT is under 400 ms and jitter is under 150 ms.
Troubleshoot Webex Calling calls in Control Hub · Hop details view, end-to-end call experience thresholds · Checked 2026-10-02
56
Webex Calling Control Hub grades an individual media relay hop as good when packet loss is under 2.5%, latency or RTT is under 200 ms and jitter is under 75 ms.
Troubleshoot Webex Calling calls in Control Hub · Hop details view, media relay point thresholds · Checked 2026-10-02

Documents

tier 1 standards and regulators

Cite this page

APA

WarmTransfer. (2026, October 2). Troubleshooting audio delay and talk-over on voice calls. WarmTransfer. https://warmtransfer.net/knowledge/call-latency-troubleshooting

BibTeX

@misc{warmtransfer-call-latency-troubleshooting,
  title  = {Troubleshooting audio delay and talk-over on voice calls},
  author = {{WarmTransfer}},
  year   = {2026},
  url    = {https://warmtransfer.net/knowledge/call-latency-troubleshooting},
  note   = {Verified 2026-10-02}
}

Troubleshooting audio delay and talk-over on voice calls

Troubleshooting audio delay and talk-over on voice calls

Delay limits and talk-over

Where the delay comes from

The de-jitter buffer

Media hops that decode and re-encode

Measuring delay with RTCP

Platform thresholds

Diagnosing it

Check where the call was anchored

Check VPN and virtual desktop paths

Pull per-call metrics

Count transcoding hops

See also

Applicability

What remains uncertain

See also

Related to

Sibling symptom

Referenced by

Sources

Documents

G.114 : One-way transmission time

RFC 3550 RTP: A Transport Protocol for Real-Time Applications

RTP Control Protocol Extended Reports (RTCP XR)

Customer network readiness (Genesys Cloud network requirements)

Edge Locations

Global Media Fabric overview

Media Quality and Network Connectivity Performance - Skype for Business Online (retired)

Network recommendations - Azure Communication Services

Plan for media bypass with Direct Routing

Prepare your organization's network for Microsoft Teams

Run the standalone Genesys Cloud WebRTC Diagnostics app

Stream classification in Call Quality Dashboard (CQD)

Troubleshoot audio quality issues by using QualityMetrics in the contact record

Troubleshoot Webex Calling calls in Control Hub

Troubleshoot your network for call quality and disconnect problems

Understanding Delay in Packet Voice Networks

Understanding Jitter in Packet Voice Networks (Cisco IOS Platforms)

Cite this page