YY Voice: the channel system that Chinese guilds grew up on
Summary
A team voice service that started in 2008 as a tool for group play and grew into a tree of permission-controlled channels. It still prioritises low latency and a fine-grained role system, with screen sharing and file transfer alongside the voice rooms.
Recommend- Name
- YY Voice
- Version
- 20260614
- Pricing
- Free
- Open source
- Proprietary
- Operating system
- WindowsmacOSAndroidiOS
- Website
- https://www.yy.com/yy8/
- Updated
- 2026-10-22 22:00
For a whole generation of players, the name of this service became a verb. You did not join a voice call, you went onto it, and everyone understood what that meant.
What voice looked like before it
Before 2008 the options for talking to a group while playing were poor. Games shipped with their own voice channels, usually compressed to the point of being hard to understand and limited to a single team. Anything more organised, a raid of twenty-five people split across roles and responsibilities, had to be run through typed chat while the fight was happening. The arrival of a standalone service with its own voice servers, multi-person channels and audio that stayed clear under load changed the way groups coordinated. Commanders could speak to subgroups, squad leaders could relay, and the whole structure held together in real time.
It began as a tool for group play and then spread sideways into music performance, online teaching and ordinary meetings, because it turned out that a low-latency room with roles was useful well beyond games. That broadening is worth remembering when judging the product today, because the current client is not obviously aimed at the players who built its reputation.

The channel tree is the real design
Everything in the product is organised around channels. A channel is an independent voice room with a complete permission model attached: administrators, sub-channels, guests, visitors, each with different rights to speak, to move, to manage and to enter. A large group might maintain dozens of sub-channels, one for commanders, several for different roles, one for substitutes waiting to be called in, one that exists purely for conversation. Moving between them is immediate, which sounds trivial until you have used a system where it is not.
The tree structure later became the standard shape for this category of software, and products built years afterwards copied the idea in a simplified form. What has not been copied as successfully is the granularity. The permission model here is fine enough to express rules like who may speak in a channel during an engagement but not between them, or who may drag another member into a different room. For a large organisation that needs enforced order in voice, that detail is not a luxury, it is the reason the tool is still in use.

Latency and stability under bad networks
Years of accumulated engineering show up in the transport layer. Voice is carried over the connectionless protocol that suits real-time audio, with packet loss compensation and a jitter buffer tuned to keep conversation intelligible when the network is not. The harder problem is the domestic one: several competing carriers whose peering between each other is uneven, so a call between two users in the same city can cross networks with very different quality. Handling that well is a local problem, and it is where a service with servers in the right places has an advantage over a foreign product that does not.
The practical measure is what two hundred milliseconds of delay does in a shooting game. Calling out a position late is the difference between taking the shot and being taken. Beyond voice, the client carries screen sharing, text chat and file transfer, and it exists on desktop and mobile alike. The honest caveat is that the desktop builds are substantially better than the mobile ones in both features and stability, which is the reverse of what most modern chat applications assume.
The turn towards broadcasting
From 2015 the centre of gravity moved. Live streaming became the commercial core, the game streaming business was eventually separated into its own platform, and the client began to open on a lobby full of live rooms with voice demoted to a secondary entry point. For people who had used it as a voice tool for a decade this felt like a rupture, and it is fair to describe it that way. It was also the predictable outcome of a business in which voice alone does not pay for the servers that voice alone requires.
Judged purely as a low-latency voice tool it remains one of the better answers, particularly for groups playing on domestic servers against domestic opponents, where its connection stability is a genuine advantage. Judged as the clean, single-purpose communication tool it once was, it is no longer that, and no amount of nostalgia will make it so.