Hello everyone,
Following the many reports from the community regarding voice communication latency, which has been perceived as too noticeable, the technical team has carried out extensive research and development. We’d like to share the results with you.
Update 1.7.7 will include a reduction in voice communication latency, although the overall latency still present in-game cannot be eliminated entirely.
For those who want more details:
Voice communication latency is standard in the context of a multiplayer video game, where data is continuously exchanged between player PCs and the game server, while VR headsets stream wirelessly to the player PCs. Voice audio must be recorded by the headset, transmitted to the player's PC, sent to the server, redistributed to the appropriate teammates' PCs, and finally streamed back wirelessly to the headsets. This process cannot happen instantaneously.
In EVA games, where players are physically standing next to one another, this latency is naturally much more noticeable than in most video games, where players are each sitting at home.
That said, this latency remains a source of frustration for the community, which is why we invested time in research and development.
To address a theory that some of you have raised from time to time: the culprit is not the headset technology!
Our measurements show that latency is very similar between Pico and Focus headsets.
The difference between the two is less than one tenth of a second, which is negligible compared to the total in-game latency, typically ranging from 0.6s to 2s depending on network conditions.
In the interest of transparency, we'd like to share the results of the improvements achieved after all the work carried out on this topic.
| Voice latency comparison before and after the changes, depending on network lag (average) | Before | After | Improvement |
|---|---|---|---|
| Standard | 0.62s | 0.41s | -0.21s |
| Simulated lag: Low | 0.81s | 0.62s | -0.21s |
| Simulated lag: Medium | 1.03s | 0.97s | -0.06s |
| Simulated lag: High | 1.07s | 1.04s | -0.03s |
| Voice latency comparison between Focus and Pico, before and after the changes, depending on network lag (average) | Pico | Focus | Pico vs Focus Difference |
|---|---|---|---|
| Before changes Standard network |
0.62s | 0.64s | -0.02s |
| After changes Standard network |
0.41s | 0.50s | -0.09s |
| After changes Simulated lag: Low |
0.62s | 0.69s | -0.07s |
| After changes Simulated lag: Medium |
0.97s | 0.95s | +0.02s |
| After changes Simulated lag: High |
1.04s | 0.98s | +0.06s |
We reduced the size of the audio packets transmitted between player PCs and the GM PC.
📦 In practical terms, when you speak, your microphone records small chunks of audio that are sent to the server and then redistributed to your teammates. Each audio packet contains a short segment of your voice. In other words, instead of sending the entire album of a sentence you've spoken, the game sends lots of singles: small recordings of around a tenth of a second as you speak.
The larger the audio packet, the better the audio quality received... but the longer the delay, because the system waits longer before sending it.
The smaller the packet, the sooner it can be sent—but the lower the perceived audio quality. This is because the audio is split into smaller chunks, making packet loss more likely.
There is therefore a trade-off between audio quality and latency. We chose to prioritize lower latency while keeping audio quality at a level we felt remained clear and understandable.
However, it's important to note that this is far from the only contributor to latency. Even if we were sending 1 ms packets, total latency would still be far from just a few milliseconds. The rest of the networking process accounts for the vast majority of the delay you hear.
We also explored several other approaches, but none of them produced meaningful improvements.
Thank you all for your patience and for taking the time to read this while we worked on the issue.
We look forward to seeing your games on version 1.7.7!
The programmers and QA team 👋🏻
Des questions ou des retours ? ⬇️