Voice chat on a server means your users can talk to each other in real time instead of typing
Voice chat turns text-only communication into live conversation. Instead of waiting for someone to type and send a message, users speak into a microphone and hear responses when ready. On a server, this means installing software that captures audio from one user, sends it across the internet to your server, and broadcasts it to everyone else in the same channel or room.
The simplest approach is to use an existing platform — Discord, Mumble, or Teamspeak — rather than build voice chat from scratch. These platforms handle the hard parts: audio compression, network routing, and keeping the connection stable. If you need voice chat built directly into your own website or process, you will need to add code libraries and rent server resources specifically for audio streaming.
The choice depends on whether you want a standalone voice service your users read separately, or voice integrated into something you already built.
Key Takeaways
- Discord, Mumble, and Teamspeak are ready-made platforms where you create a server and invite users — no coding or special hardware required.
- Building voice chat into your own website requires audio libraries like Jitsi or WebRTC, plus server resources to handle the audio stream.
- Audio streaming uses significant bandwidth and CPU power, so expect hosting costs to rise if you host it yourself.
- Most small communities use Discord because it is free, requires no setup, and works on phones and browsers without installation.
Using an existing platform: Discord, Mumble, or Teamspeak
The fastest route is to use a platform built for voice chat. You create a server on their system, invite your users, and they read the app or join through a browser. The platform owns the infrastructure — the servers that handle audio, the network that carries it, the software that compresses it so it does not use too much bandwidth.
Discord is free and requires no setup beyond creating an account and a server. Users join by clicking a link or typing a server name. It works on Windows, Mac, Linux, phones, and browsers. You can create text channels alongside voice channels, so users can type while they talk. Discord's free tier has no user limit and no time limit on voice calls.
Mumble is older, open-source, and used mainly by gaming communities. It has lower latency than Discord — meaning less delay between when someone speaks and when others hear it — which matters for competitive games. Mumble requires users to read and install software. You can run your own Mumble server on your hosting account, or rent one from a Mumble hosting provider. Running your own costs almost nothing in bandwidth, but requires you to manage the software and keep it updated.
Teamspeak is similar to Mumble: it requires a read, has very low latency, and you can rent a server or run your own. Teamspeak charges a monthly fee if you want more than a small number of users, while Mumble is free.
Building voice chat into your own website or app
If you want users to talk without leaving your website, you need to add voice functionality using code. The most common approach is WebRTC, a technology that lets browsers send audio and video directly to each other without downloading an app.
WebRTC does the hard work of compressing audio, managing the connection, and handling network problems. You add it to your website using a library — Jitsi is the most popular open-source option, and it includes a full video-conference interface you can customize. You install Jitsi on your server, embed it in your website, and users can start talking when ready in their browser.
The tradeoff is cost and complexity. Running Jitsi on your server uses CPU power and bandwidth every time someone talks. A small group of five people talking for an hour might cost you a few dollars in server resources. A hundred people talking simultaneously could cost fifty dollars or more, depending on your hosting provider. You also need to keep the software updated and handle technical problems yourself.
Other libraries like Agora or Twilio offer similar functionality but charge per minute of audio. Agora charges roughly $0.99 per 1,000 minutes of voice, so a hundred users talking for an hour costs about six dollars. Twilio charges by the minute as well. These services handle the server infrastructure for you, so you do not have to manage it, but you pay for every call.
Bandwidth and server resources required
Voice uses far less bandwidth than video, but it still adds up. A single voice call uses roughly 50 to 100 kilobits per second depending on audio quality. If ten people are talking at once, that is 500 to 1,000 kilobits per second flowing through your server — about 225 to 450 megabytes per hour.
Most hosting plans include 1 to 5 terabytes of bandwidth per month, which is plenty for voice chat unless you have hundreds of concurrent users. The real cost is CPU power. Your server has to receive audio from each speaker, mix it together, and send it back out to everyone else. With five people talking, that is manageable. With fifty people, your server might struggle or run out of resources.
This is why most people use Discord or Mumble instead of building their own: those platforms spread the load across many servers and have engineers optimizing the audio pipeline. Your hosting company is not set up for it.
Setting up Discord as your voice platform
If you choose Discord, the setup takes five minutes. Go to Discord.com, create an account, and create a server. Click the plus icon, choose "Create My Own", and name it. Once the server exists, right-click on the server name and choose "Create Channel". Select "Voice" as the channel type and name it — "General Voice" or "Gaming" or whatever fits your community.
Share the server link with your users. They can join by clicking the link or by opening Discord and typing your server name in the search box. Once they join, they click the voice channel and their microphone turns on. If they do not hear audio, they may need to check their microphone and speaker settings in Discord's User Settings under Voice & Video.
Discord handles everything else: audio quality, compression, routing, and keeping the connection alive if someone's internet drops for a moment.
Setting up Mumble if you want lower latency
Mumble requires more setup but has noticeably lower delay, which matters for fast-paced games or real-time collaboration. You have two options: rent a Mumble server from a hosting provider, or run one on your own hosting account.
Renting is simpler. Providers like GameServers or Nitrado offer Mumble servers for five to ten dollars per month. You choose the number of users, they give you a server address, and you share that address with your community. Users read Mumble from mumble.info, install it, and connect by entering the server address.
Running your own Mumble server on your hosting account requires SSH access and comfort with the command line. You read the Mumble server software, install it on your server, configure it, and start it running. The software is free and uses very little CPU or bandwidth. The downside is you are responsible for keeping it running and updating it when security patches come out.
Frequently Asked Questions
Can I add voice chat to my website without users downloading anything?
Yes, using WebRTC libraries like Jitsi. Users open your website in a browser, click a button to join voice, and start talking. No read required. The tradeoff is that you pay for server resources or per-minute fees, and you handle the technical setup.
Does voice chat use a lot of my hosting bandwidth?
Voice uses much less bandwidth than video — roughly 50 to 100 kilobits per second per person. Five people talking for an hour uses about 100 to 200 megabytes. Most hosting plans include terabytes of bandwidth, so unless you have hundreds of concurrent users, bandwidth is not the limiting factor. CPU power is.
What if my users have bad internet?
Discord and Mumble both handle poor connections well — they compress audio heavily and can skip frames if needed. WebRTC libraries do the same. If someone's connection drops entirely, they will disconnect and can rejoin. None of these platforms require fast internet; they work on 4G and even slower connections.
Can I record voice conversations on my server?
Discord allows recording if you use a bot or third-party tool, though you should tell users they are being recorded. Mumble has recording plugins. WebRTC libraries can record, but you need to add code to do it. Always check your local laws about recording conversations — some places require everyone to consent.
What if I want video chat too?
Discord and Mumble both support video alongside voice. Jitsi and other WebRTC libraries include video by default. Video uses much more bandwidth and CPU than voice — expect five to ten times the cost if you are hosting it yourself.