Livestreaming a DJ Set: Audio, Encoders, OBS and Latency.

Broadcasting a DJ set live is not a matter of pressing the start button and hoping for the best. Almost every problem you see on a channel—flat sound, out-of-sync speech, dropped frames or clipping—starts before the software opens, in the way the signal leaves the booth, reaches the encoder and uploads to the platform.

This guide follows that complete chain from left to right: which signal to take from the mixer, when you need an external interface, how to organise routing to the PC, what OBS offers compared with Streamlabs, how audio/video latency and lip sync fit together, and which encoder settings support a stable broadcast.

From Booth to Encoder

A broadcast has six stages, and each fails in its own way:

1. Booth. The mixer produces the mix. This is where you decide which signal leaves: master, booth or the software’s own output.
2. Interface and routing. The signal enters the computer through an external interface, the controller’s USB output or a virtual audio device.
3. Broadcast software. OBS, Streamlabs or a similar application combines your audio and video sources into scenes.
4. Encoder. It compresses the result into video and audio codecs, typically H.264 and AAC, at a constant bitrate.
5. Platform. It receives the stream over RTMP or RTMPS, transcodes it and distributes it through its CDN.
6. Viewer’s player. It stores a buffer before playback; that buffer is the last part of latency you do not control.

The principle is simple: correct each problem at the stage where it occurs. If sound already arrives saturated from the booth, no OBS filter will repair it; if upload cannot sustain the bitrate, no latency option will save the broadcast.

Internal Audio vs Master Output

Your first decision is which audio the software collects. There are two paths:

• Internal audio. This is sound produced by the computer itself: the controller’s output when it acts as a soundcard, the music application or the player. OBS captures it as desktop audio or an output device. It is the shortest path and, in many setups, the only possible one when the controller exposes just one USB channel.
• Master output. This is the stereo signal leaving the physical mixer, with gain already set in the booth, channel EQ applied and the crossfader where you left it. It reaches the software through an external interface and gives you precise control over what the viewer hears.

Which should you use? If you will speak over the music, mix two external sources or want level to be independent of Windows or macOS, use master output. If you broadcast directly from software and its virtual master is your only source, internal audio is perfectly valid. Do not combine both at once: you will duplicate the signal and reach the viewer with echo or twice the level.

One often-overlooked detail: headphone output is not master output. If you capture what you hear, the audience receives the monitoring bus with its cuts and working volume, rather than the mix.

Routing: External Interface and Output to the PC

Routing means deciding which signal travels along each path. In a broadcast, separate three routes: what plays through your speakers, what the viewer hears and what gets recorded.

External interface. It is worthwhile when the laptop output overloads at booth levels, when you need balanced XLR or TRS microphone inputs, or when you want low-latency measurement and monitoring. The setup is straightforward: connect the mixer’s balanced master to the interface inputs and select that interface as the capture device in your software. Before wiring the booth, review input and output cabling.

Virtual devices. Tools such as a virtual audio cable create internal paths: for example, sending music to broadcast software while you hear something else through the physical interface. They help separate routes, but multiply failure points: every duplicated capture is a potential echo.

Gain. Set level in the booth rather than the software. At the mixer, seek margin for peaks; in software, adjust only when there is a genuine level difference between sources. This is the same working order used with dBFS and LUFS meters.

OBS, Streamlabs and Alternatives

OBS is the open-source standard: scenes with sources, a mixer with per-source meters, audio filters for gain, noise suppression and delay, and outputs for both broadcasting and recording. Its auto-configuration wizard suggests starting values according to your hardware and connection, and the official recommendation is to test with a recording before broadcasting.

Streamlabs and alternatives. Streamlabs Desktop starts from the same idea and adds alert widgets, ready-made sources and per-application capture; it also imports configuration if you previously used OBS. Common alternatives include programmes with virtual studios and simple camera switching, useful if you produce a lot of content with guests. The choice is not ideological:

• If you want the smallest maintainable setup and full control over every filter, choose OBS.
• If you want alerts, overlays and social panels without building them, choose Streamlabs.
• If your computer is a Mac and you want native multicamera switching, examine macOS alternatives before installing anything.

In all three cases, audio follows the same flow described above: the correct input device, active meters and one capture per signal.

Audio/Video Latency and Lip Sync

Audio/video latency. This is the delay between something happening in your booth and the viewer seeing it. It accumulates at four points: the encoder buffer, the network, platform processing and the player’s buffer. Platforms offer normal, low and ultra-low latency modes; lowering latency reduces the playback buffer and leaves less margin for network variations. The right mode therefore depends on whether you will talk with chat or simply broadcast music.

Lip sync. When audio arrives after the image, the usual causes are camera processing, HDMI capture-device delay, delayed audio filters or missing capture timestamps. The correct method is to record a clap or clapperboard at startup, measure the offset in the recording and correct it through the audio source’s sync offset or video delay. Correcting by ear while the set plays almost always makes the result worse.

Keep these two discussions separate: latency affects interaction with the audience; synchronisation affects the image’s credibility. You can live with three seconds of latency; with half a second of lip-sync offset, you cannot.

Microphone and Multicamera Capture

Microphone. For speaking in a venue with music, a cardioid dynamic microphone handles the room better than a sensitive condenser. Sing away from the capsule, work off-axis and raise gain to a healthy level, with peaks far below the ceiling. Keep noise suppression gentle: an aggressive filter eats the voice when volume falls. Avoid monitoring through speakers while speaking, because the microphone returns to the room. Technique details are in microphone technique, vocals and talkover.

Multicamera capture. Each camera is another source: a wide booth shot, a close-up of hands or a view of the crowd. Practical rules: use one capture device per camera when connecting over HDMI, keep every stream at the same frame rate, and distribute USB devices over different controllers to avoid saturating the bus. If upload cannot keep up, reduce resolution before reducing the number of cameras: two stable shots are worth more than four that cut out.

Encoder: Bitrate, Codec and Stable Upload

The settings that sustain a broadcast are short and specific:

• Video codec: H.264 is compatible with almost everything.
• Encoding: CBR (constant bitrate). Live stability matters more than momentary efficiency.
• Keyframe: every 2 seconds, as platforms recommend; at most, 4.
• Audio: stereo AAC at 128 kbps, with a 44.1 kHz or 48 kHz sample rate.
• Video bitrate: for 1080p at 30 fps with H.264, YouTube’s reference is a minimum of 5 Mbps and around 14 Mbps recommended; for 720p at 30 fps, 3 Mbps and 8 Mbps.
• Upload margin: do not occupy your entire connection. Leave room for network variation and the extra upload needed when broadcasting directly to several destinations.

Before broadcasting, test with audio and movement similar to the real conditions, and watch the broadcast status during its first minutes. If dropped frames appear, act in this order: lower bitrate, lower resolution, switch to hardware encoding and, only at the end, reduce the number of sources.

Next Step

With the broadcast chain set up, the next step is giving the channel content: how to structure a music radio programme or podcast with jingles, beds and broadcast loudness. Once you broadcast every week, organise the operation of platforms, copyright, recording and backup.