Set Up a Server for IP Camera Streams, Including Audio

The short answer

An IP camera server is software that pulls the camera's network stream, usually over RTSP, and passes it on to the programs that need it. It can repackage the stream into formats such as WebRTC or HLS, or feed it to a recording or virtual-camera tool. It is often called a media server, relay or restreamer. You can run it on a home computer, a small single-board computer or a NAS that supports it. For speech capture, one condition matters most. Video routing is well supported, but audio often has to be handled separately. It only works if the camera has a microphone, the camera's audio is turned on, and the audio codec is one the next step accepts.

No device-specific evidence was supplied for this guide: no camera models, measurements or test results. Everything below is based on general, verifiable behavior of the protocols and common tools. Confirm the details for your own camera in its documentation or settings.

How an IP camera setup differs from a USB webcam

A USB webcam with a built-in microphone appears to the operating system as two devices, a camera and an audio input. Conferencing apps can select both directly. An IP camera appears as neither. It is a network device that serves a compressed stream. Your computer has to receive that stream, decode it and then present the video and audio to other software. The stream server and any virtual-device tools do that work, and each stage is a place where the audio can be dropped, delayed or rejected.

What the server chain typically looks like

• 1. The camera publishes a stream, most commonly RTSP. Many cameras also support ONVIF, a standard for discovering the camera and finding its stream addresses.
• 2. A media server or relay connects to that stream once and makes it available to several consumers. This avoids each app opening its own connection to the camera.
• 3. The server repackages the stream for each destination. For example, it may serve WebRTC for low-latency browser viewing, HLS for broader playback compatibility, or keep RTSP for recording software.
• 4. To use the camera in a conferencing app, a broadcasting tool such as OBS Studio can take the stream as a source and expose it as a virtual camera.
• 5. Audio is routed to the conferencing app as a microphone input. This usually happens through a separate path, as explained below.

Open-source relays such as MediaMTX and go2rtc, and the command-line tool FFmpeg, are widely used for steps 2 and 3. Network video recorder software can also act as the server. Which one fits depends on your hardware and the output formats you need. This guide does not compare them, because no comparative evidence was supplied.

Where speech audio tends to get lost

Not every IP camera has a microphone. Some only have an audio input jack for an external mic. On cameras that do have audio, it is sometimes switched off by default in the camera's own settings. Check this first, because no server configuration can recover audio the camera never sends.

Codec compatibility is the next hurdle. A codec is the method used to compress the audio. IP cameras commonly send G.711, a basic telephone-grade codec, or AAC, a codec more common in general media. Some cameras offer other options. WebRTC in browsers supports Opus and G.711 but not AAC. HLS playback normally expects AAC. If the camera's codec does not match the output format, the server must transcode the audio, meaning it converts it to another codec. Otherwise the audio is silently omitted while the video plays normally. Many relays can transcode audio by calling FFmpeg, but this has to be set up deliberately.

The virtual camera step is a common gap. A virtual camera carries video only. The conferencing app still needs a separate microphone input. One common approach is to send the stream's audio from your broadcasting tool to a virtual audio device, then select that device as the microphone in the meeting app. Whether this works smoothly depends on your operating system and tools. It is worth testing before relying on it.

Check whether the stream carries audio

Before configuring a server, confirm what the camera actually sends. FFmpeg's ffprobe tool lists the streams inside a feed. Put your camera's RTSP address, including any credentials, in a shell variable rather than typing it in plain view. Then run:

ffprobe -rtsp_transport tcp -i "$CAMERA_RTSP"

The output lists each stream. If there is no line marked Audio, the camera is not sending sound, so check its settings or hardware. If an audio stream is listed, note the codec name and compare it with the codecs supported by your planned output format.

Practical implications for voice quality

• Sound quality: A camera microphone is usually placed for monitoring a room, not for close speech pickup. Telephone-grade codecs such as G.711 also limit how full a voice can sound. The supplied evidence does not show how any specific camera performs, so judge by test recordings at your normal speaking distance.
• Sync and delay: Video and audio pass through buffering and possibly transcoding. They can drift apart or add noticeable delay. Low-latency outputs such as WebRTC help, and broadcasting tools usually let you offset the audio to restore lip sync.
• Reliability: Each extra stage, from relay to transcoder to virtual audio device, is another place that can fail after a reboot or network change. Fewer stages are easier to maintain.
• The simpler alternative: If clear speech in calls is the goal, a USB microphone or headset alongside the IP camera video often avoids the audio routing problem entirely. The camera then only needs to supply the picture.

Next step

Start by confirming that your camera sends an audio stream and identifying its codec. If it does, pick a relay that can output a format your viewing or conferencing path accepts, and transcode the audio if needed. Then route the audio to an input your meeting app can select. If the camera has no usable audio, plan a separate microphone from the start rather than building the server around sound it cannot deliver.