R.E.P.O., a co-op horror game whose Steam page sells “proximity voice chat”, peaked at 266,908 concurrent players, yet the only direct adoption figure for spatial voice we could find is 17 years old: 50% of Second Life users in 2009 (SteamCharts, R.E.P.O. 2026; Linden Lab 2009, via Wadley, Carter and Gibbs). Proximity voice (also called positional or spatial voice) makes a player’s volume depend on how close their character is, so a game world behaves like a room. The surrounding data is thin but useful: 59% of U.S. teens who played online with others used a voice connection in 2015 (Pew Research Center, Teens, Technology and Friendships 2015), and 25% of multiplayer players often or always voice chat with their whole team (Unity, multiplayer gamer study). Roblox handles millions of minutes of voice daily on a system that is proximity-based by default (Roblox, Voice Safety Classifier v3 2026). We aggregated data from Pew Research Center, VRChat, Ofcom, the U.S. Federal Trade Commission, and peer-reviewed work in the journal Human-Computer Interaction, and we say plainly below what nobody measures. For harm and moderation rates across all voice channels, see our voice chat toxicity statistics.
TL;DR
- No public survey we found measures the share of players who use proximity voice in games; Valve’s Steam Hardware Survey has no voice or microphone category (Valve, Steam Hardware Survey September 2026).
- 50% of Second Life users had adopted its spatial voice channel two years after launch (Linden Lab 2009, via Wadley, Carter and Gibbs).
- 59% of U.S. teens who play online with others used a voice connection, 71% of boys against 28% of girls (Pew Research Center 2015).
- 25% of multiplayer players often or always voice chat with their whole team, against 33% who text chat (Unity, multiplayer gamer study).
- R.E.P.O. peaked at 266,908 concurrent Steam players and PEAK at 169,910 (SteamCharts, October 2026 snapshot).
- Vivox defaults to a 32 meter hearing range, while VRChat’s player voice defaults to 25 meters (Unity Vivox documentation; VRChat Creator Docs).
- The game-specific proximity evidence we could read rests on just 27 trial and interview participants (our sum of 9, 8 and 10, section 4).
- 55% of UK child online gamers play at least sometimes with people they have never met (Ofcom, Growing Up Gaming 2026).
- Epic paid a $275 million COPPA penalty in a case that also changed Fortnite’s voice chat defaults for minors (FTC 2022).
- Roblox’s voice classifier reaches 61% recall at a 1% false positive rate (Roblox, Voice Safety Classifier v3 2026).
1. Voice Chat Use: The Baseline Proximity Sits On
Proximity voice is a subset of voice chat, and even the parent number is shaky. Only 25% of players often or always voice chat with their whole team, so the open-mic image of multiplayer games describes a minority habit, not the norm. The question for a designer is not whether players own a microphone but whether they will speak to whoever happens to be nearby.
The gender split is the number that has aged least comfortably: at 71% against 28%, voice was already a boys’ channel before proximity became a design trend. The 2024 teen gaming page from the same researchers reports no voice chat figure, so the 2015 report is the most recent available data from that series: Pew Research Center, 2015.
| Metric | Value | Source |
|---|---|---|
| U.S. teens who play online with others and use a voice connection | 59% | Pew Research Center, Teens, Technology and Friendships 2015 |
| Boys vs girls among those teens using a voice connection | 71% vs 28% | Pew Research Center 2015 |
| Boys aged 15 to 17 using a voice connection | 75% | Pew Research Center 2015 |
| Players who often or always chat with their whole team, voice vs text | 25% vs 33% | Unity, in-game chat guide (survey year not stated) |
| Gamers who cite in-game chat as a key factor in choosing their next game | 31% | Unity Multiplayer Report 2022, via 80 Level |
| Chinese players who use built-in voice chat when playing, ever vs often | 90.6% vs 38.4% | Tencent Cloud’s own research (vendor claim), via HackerNoon |
| Those Chinese players who move to a third-party service when a game lacks voice | 73.7% | Tencent Cloud’s own research (vendor claim), via HackerNoon |
| Self-identified U.S. teen gamers vs all U.S. teens who use Discord | 44% vs 28% | Pew Research Center, Teens and Video Games Today 2024 |
Context note: the 90.6% and 25% figures do not conflict so much as ask different questions, one about ever using voice and the other about habitually using it with everyone on the team. Tencent Cloud also sells voice services, and we could not read its whitepaper directly. The same developer guide adds a vendor claim that players using voice spent twice as much time in the same game as those who did not, but that is a correlation from Vivox, a voice provider. The Entertainment Software Association’s survey of 24,216 active players in 22 countries reports that players under 35 use voice chat more often, with no percentage in the readable summary. The third-party habit is why voice tools such as a voice changer for Discord sit outside the game client.
2. Proximity Titles on Steam: Demand Without a Usage Rate
Concurrent players (people inside a game at the same moment) are the best public signal available, and it is a proxy. R.E.P.O. reached 266,908 concurrent players on a pitch that names proximity voice chat directly, and PEAK’s Steam page lists “Communicate through proximity chat” as a feature. That tells us studios now sell the mechanic on the box. It does not tell us how many of those players had a microphone on.
Longevity is the more interesting signal. DayZ’s October 2026 peak of 69,704 sits at 88.5% of its all-time high (69,704 / 78,739), while R.E.P.O.’s October peak of 71,168 is 26.7% of its record (71,168 / 266,908). A hit can fade fast, and an older survival game with proximity voice can hold its audience. Steam counts come from SteamCharts, an independent tracker of Steam’s public player data; R.E.P.O.’s store page showed 140,942 English reviews at 96% positive.
| Metric | Value | Source |
|---|---|---|
| R.E.P.O. all-time peak concurrent players on Steam | 266,908 | SteamCharts, R.E.P.O. |
| PEAK all-time peak concurrent players on Steam | 169,910 (July 2025) | SteamCharts, PEAK |
| Phasmophobia all-time peak concurrent players on Steam | 112,241 | SteamCharts, Phasmophobia |
| DayZ all-time peak concurrent players on Steam | 78,739 | SteamCharts, DayZ |
| Hell Let Loose all-time peak concurrent players on Steam | 21,086 (December 2024) | SteamCharts, Hell Let Loose |
| DayZ average vs peak concurrent players, October 2026 | 34,388.14 vs 69,704 | SteamCharts, DayZ |
| PEAK English-language Steam reviews and share positive | 147,990 reviews, 95% positive | Valve, PEAK Steam store page, October 2026 |
| Second Life users who had adopted spatial voice, two years after it arrived | 50% | Linden Lab 2009, via Wadley, Carter and Gibbs, Human-Computer Interaction |
Context note: proximity status comes from store text for R.E.P.O., PEAK and Hell Let Loose (“In-game proximity voip, command voip and unit voip”), from peer-reviewed work for DayZ, and from a 2026 study for Phasmophobia. Second Life’s figure is the most recent available direct adoption share: Linden Lab 2009. For how adjacent worlds fare, see our social VR statistics.
3. How Far a Voice Carries: Documented Audible Ranges
Audible range is the one part of this topic with hard numbers, because engineers have to write them down. DayZ’s 50 meter voice range was double VRChat’s 25 meter default, and because audible area grows with the square of the radius, that doubling means four times the ground ((50 / 25) squared = 4). Range is therefore a social decision: it sets how many strangers can overhear a sentence.
The other design tell is the gap between voice and everything else. VRChat lets avatar sounds carry to 40 meters by default, 1.6 times its voice range, while VRChat’s documentation tells world builders to leave the near distance at zero. Vivox’s guidance goes further, giving a formula so voices fade to near zero at the edge of the range instead of cutting off abruptly.
| Metric | Value | Source |
|---|---|---|
| Vivox default maximum hearing distance (AudibleDistance) | 32 meters (about 105 feet) | Unity, Vivox positional channel properties |
| Vivox default distance kept at full volume (ConversationalDistance) | 1 meter (about 3 feet) | Unity, Vivox positional channel properties |
| VRChat default distance where a player’s voice ends (Distance Far) | 25 meters | VRChat Creator Docs, Player Audio |
| VRChat default voice gain, and the allowed range | 15 dB, range 0 to 24 dB | VRChat Creator Docs, Player Audio |
| VRChat default range for avatar sounds (Far Radius) | 40 meters | VRChat Creator Docs, Player Audio |
| DayZ range for voice between avatars, as documented in 2015 | 50 meters | Wadley, Carter and Gibbs, Human-Computer Interaction |
Context note: these are configuration defaults, not measurements of what players hear, and game units may not match real meters. Epic says Fortnite creators can set the distance at which volume drops to zero, but its announcement gives no value in meters. Perceived direction also depends on the listener’s headset and processing, which our headset localization statistics cover in detail.
4. What Research Says Proximity Changes
The research direction is consistent and the sample sizes are tiny. The game-specific proximity studies we could read total 27 participants (9 + 8 + 10), which is enough to generate good hypotheses and not enough to estimate anything for a population. The clearest finding comes from Wadley, Carter and Gibbs: players in a team shooter first saw spatial voice as a restriction, then “discovered advantages of the spatial mode and used the radio mode less often”. Realism was accepted only if it helped, or at least did not hinder, winning.
DayZ shows the social consequence. Because voice was stereo-directional, speaking revealed position, and “non-participation in voice-chat was sometimes interpreted as an act of hostility”. Silence stops being neutral once a stranger can hear you. A 2026 Uppsala University study reached a compatible result in newer games: proximity kept players physically closer and raised immersion, while standard voice over Discord made coordination easier.
| Metric | Value | Source |
|---|---|---|
| Players in the spatial voice trial in Wolfenstein: Enemy Territory | 9 men aged 20 to 40 | Wadley, Carter and Gibbs, Human-Computer Interaction |
| Length of that spatial voice trial | 4 weeks, 5 play sessions | Wadley, Carter and Gibbs |
| Evidence base of the DayZ spatial voice study | 60+ videos and forum stories, 8 player interviews | Wadley, Carter and Gibbs |
| DayZ survey respondents who identified as female | Less than 1% of 1,800 | Wadley, Carter and Gibbs |
| Participants who played R.E.P.O. and were then interviewed on proximity chat | 10 | Swedish study, DiVA repository 2026 |
| Games compared with proximity chat vs Discord voice, each played twice by the author | 3 (PEAK, R.E.P.O., Phasmophobia) | Uppsala University repository 2026 |
| Technology company employees in a spatial audio on vs off experiment (video meetings, not games) | 75 employees, 2 group tasks each | Nowak et al., CHIWORK 2023 |
| Combined participants across the three multi-person game studies above | 27 (9 + 8 + 10) | Our sum of the rows above |
Context note: the Microsoft-led meeting study found spatial audio raised perceived interactivity, shared space and ease of understanding, but its effects came from work calls and may not transfer to game worlds. One 2025 network analysis in the Proceedings of the ACM on Human-Computer Interaction found clans in a tank-based multiplayer game that use voice during battle are rated significantly higher, though the abstract gives no effect size and does not study proximity voice. The DayZ survey figure comes from a survey the authors say was not part of their analysis, and the DayZ data dates from 2012 and 2013.
5. Strangers in Earshot: Children, Contact, and Default Settings
Proximity voice has a property party chat lacks: by design it connects people who did not choose each other. That makes stranger contact the relevant risk measure. 55% of UK child online gamers play at least sometimes with people they have never met, and 64% of their parents name talking to strangers as a concern, which puts it ahead of bullying (Ofcom, Growing Up Gaming 2026). Yet rules on who a child can play with are the ones parents set least often, at 43%.
Regulators have already acted on defaults. The FTC’s 2022 action said Epic employees urged the company as early as 2017 to require users to opt in for voice chat, and the order requires voice and text chat to be off by default for children and teens. Epic’s 2024 announcement, dated August 6, says proximity chat is off by default on islands and follows parental voice settings. Default-off is now the safe design pattern, and a stranger contact measure still says nothing about how often voice is the channel used.
| Metric | Value | Source |
|---|---|---|
| UK child online gamers (aged 3 to 17) who play at least sometimes with people they have never met | 55% | Ofcom, Growing Up Gaming 2026, via LBC |
| UK child online gamers who play with such people all or most of the time | 20% | Ofcom, Growing Up Gaming 2026, via LBC |
| UK parents of gamers concerned about strangers through games or chat vs about bullying | 64% vs 60% | Ofcom, Growing Up Gaming 2026, via LBC |
| UK parents of gamers who set rules on who a child plays with vs on time limits | 43% vs 56% | Ofcom, Growing Up Gaming 2026, via BroadbandSwitch |
| U.S. teens aged 13 to 17 playing multiplayer in a typical week with strangers vs friends or family | 66% vs 72% | Digital Wellness Lab, Pulse Survey 2024 |
| U.S. teens with frequent real-time conversations in multiplayer, with strangers vs friends or family | 32% vs 54% | Digital Wellness Lab, Pulse Survey 2024 |
| U.S. teen players harassed or bullied while playing, in at least one of three measured ways | 43% | Pew Research Center, Teens and Video Games Today 2024 |
| Civil penalty Epic agreed to pay for alleged children’s privacy violations | $275 million | FTC, Epic Games settlement 2022 |
Context note: Ofcom’s base was 2,522 parents and 1,719 children, with parents answering for children aged 7 or younger, as reported by LBC and BroadbandSwitch. None of these surveys isolates proximity voice. The Digital Wellness Lab’s “real-time conversations” may include text or other channels.
6. Roblox: Safety Data From a Proximity-First Platform
Roblox is the closest thing the industry has to a large-scale proximity voice dataset, because its creator documentation says voice is proximity-based by default. It launched as Spatial Voice in September 2021, first with about 5,000 developers. U.S. abuse reports fell by over 50% per hour of speech after Roblox’s first classifier and user notifications shipped (Roblox, Voice Safety Classifier v3 2026). That is a company figure on its own product, and it is still the strongest outcome number we found for voice heard by distance.
Recall here means the share of real violations a system catches. At 61%, the other 39% (100 - 61) pass at that setting, so automated review is a filter, not a guarantee. The company’s 2026 post and 2024 post never separate proximity conversations from other voice traffic, so none of this reads as a proximity-specific rate. Call of Duty rows sit below for scale, and the figures we read for that game do not isolate proximity conversations either.
| Metric | Value | Source |
|---|---|---|
| Roblox voice default mode and who can use it | Proximity-based, age-verified users aged 13+ who opt in | Roblox Creator Docs, Voice Chat |
| Maximum server size for places using Roblox voice chat | 100 players | Roblox Creator Docs, Voice Chat |
| Roblox voice data handled | Millions of minutes daily, across 30 languages | Roblox, Voice Safety Classifier v3, June 17, 2026 |
| Fall in severe voice abuse reports vs fall in violations per minute of speech, after the 2024 model | 15.3% vs 11.4% | Roblox, Deploying ML for Voice Safety 2024 |
| Voice classifier v3 recall at a 1% false positive rate, weighted by Roblox language mix | 61% | Roblox, Voice Safety Classifier v3 2026 |
| Call of Duty accounts that faced enforcement for disruptive voice chat, August 2023 to January 2024 | 2+ million | Activision, January 2024 report, via Decrypt |
| Call of Duty players who reported the disruptive voice chat they encountered | About 1 in 5 | Activision, January 2024 report, via Decrypt |
Context note: every platform figure above is the company’s own reporting, and we found no independent audit of it.
Summary: In-Game Proximity Voice Chat Usage by the Numbers
| Metric | Value | Source |
|---|---|---|
| Second Life users who had adopted spatial voice by 2009 | 50% | Linden Lab 2009, via Wadley, Carter and Gibbs |
| U.S. teens who play online with others and use a voice connection | 59% | Pew Research Center 2015 |
| Boys vs girls among those teens using a voice connection | 71% vs 28% | Pew Research Center 2015 |
| Players who often or always chat with their whole team, voice vs text | 25% vs 33% | Unity, in-game chat guide |
| R.E.P.O. all-time peak concurrent players on Steam | 266,908 | SteamCharts |
| PEAK all-time peak concurrent players on Steam | 169,910 | SteamCharts |
| Vivox default hearing distance vs full-volume distance | 32 meters vs 1 meter | Unity, Vivox documentation |
| VRChat default player voice range | 25 meters | VRChat Creator Docs |
| DayZ voice range, as documented in 2015 | 50 meters | Wadley, Carter and Gibbs |
| Participants across three multi-person game proximity studies | 27 | Our sum, section 4 |
| UK child online gamers who play at least sometimes with people never met | 55% | Ofcom, Growing Up Gaming 2026 |
| UK parents concerned about strangers through games or chat | 64% | Ofcom, Growing Up Gaming 2026 |
| U.S. teens playing multiplayer with strangers in a typical week | 66% | Digital Wellness Lab 2024 |
| Civil penalty Epic agreed to pay in the COPPA case | $275 million | FTC 2022 |
| Roblox voice classifier v3 recall at a 1% false positive rate | 61% | Roblox 2026 |
| Fall in U.S. Roblox abuse reports per hour of speech after first classifier | Over 50% | Roblox 2026 |
Methodology and Sources
We searched for direct measures of proximity voice use, then widened to in-game voice use, stranger contact and platform documentation. Each source below was read during this update, except where the Data watch note says otherwise.
- Pew Research Center, Teens, Technology and Friendships 2015
- Pew Research Center, Teens and Video Games Today 2024
- Unity, in-game chat guide for developers
- Unity Multiplayer Report 2022, via 80 Level
- Unity, Vivox positional channel properties
- VRChat Creator Docs, Player Audio
- Wadley, Carter and Gibbs, Voice in Virtual Worlds, Human-Computer Interaction
- Nowak, Tankelevitch, Tang and Rintel, Hear We Are, CHIWORK 2023
- Uppsala University repository, An analysis on proximity chat in games and the effects it has on the players, 2026
- Swedish study of proximity voice chat in R.E.P.O., DiVA repository 2026
- Bisberg et al., Communication Patterns Predict Team Skill in Multiplayer Online Games, 2025
- SteamCharts player data and Valve store pages for R.E.P.O., PEAK and Hell Let Loose
- Valve, Steam Hardware and Software Survey, September 2026
- Ofcom, Growing Up Gaming 2026, via LBC and via BroadbandSwitch
- Digital Wellness Lab, Pulse Survey on Digital Gaming and Social Interaction 2024
- Federal Trade Commission, Epic Games settlement 2022
- Epic Games, Introducing Proximity Chat 2024
- Roblox Creator Docs, Voice Chat, Voice Safety Classifier v3 2026 and Deploying ML for Voice Safety 2024
- TechCrunch on Roblox Spatial Voice, September 21, 2021
- Tencent Cloud research, via HackerNoon
- Activision January 2024 report, via Decrypt
- Entertainment Software Association, Global Video Games Report 2025, via GameDev Reports
- Data watch: nothing we read measures the share of sessions, players or minutes spent on proximity voice; the Steam Hardware Survey has no voice category, and the Newzoo and ESA material we reached gives no percentage for it. Most recent available data flags apply to Pew 2015, Linden Lab 2009, and the Wadley, Carter and Gibbs studies (publisher date May 2015; a University of Melbourne listing says 2014). Ofcom’s own pages blocked our reader, so its figures were read via LBC and BroadbandSwitch and should be confirmed against the Ofcom summary published October 13, 2026. Activision’s original post also failed to load, so its figures come from Decrypt, and the two Swedish studies were read as abstracts only, so their degree level is unconfirmed. The Tencent Cloud whitepaper page loaded without its text, and its figures are undated. The Digital Wellness Lab describes over 1,400 adolescents, while the Ruderman Family Foundation’s release on the same work says about 1,500 and rounds the stranger figure to nearly 70%. Lethal Company (SteamCharts all-time peak 239,369) is left out of section 2 because its store page does not mention proximity chat and we could not confirm it from a primary source.
Last updated: October 31, 2026. We update this roundup quarterly, and the next refresh is expected when Ofcom publishes its next Children and Parents: Media Use and Attitudes report.