HLS is how every viewer watches your stream on Twitch, YouTube, and Kick. Understanding it explains why StreamTranslate's overlay approach to captions is the most reliable choice for any stream setup.
Try StreamTranslate FreeHTTP Live Streaming (HLS) is a streaming protocol developed by Apple and published as an Internet standard. It's the delivery mechanism that takes your stream from the platform's servers and gets it to viewers' browsers, phones, and TVs. Virtually every major streaming platform — Twitch, YouTube Live, Facebook Gaming, Kick, and hundreds of others — uses HLS to deliver streams to viewers.
HLS works by breaking the live stream into small video segments (typically 2-10 seconds each) and serving them over standard HTTP connections. A manifest file (the .m3u8 playlist) tells the player which segments to download in sequence. This segment-based approach has a major advantage over older streaming protocols: it works over standard web infrastructure (CDNs, caches, load balancers) that HTTP traffic already uses efficiently.
The main tradeoff of HLS is latency. Each segment must be created, uploaded, and cached before viewers can download it, adding 10-30 seconds of delay in standard mode. Low-Latency HLS (LL-HLS) reduces this by using partial segments and push-based delivery, getting latency down to 2-5 seconds on supported platforms and networks.
StreamTranslate operates upstream of HLS delivery — it generates caption overlays that are composited into your video by OBS before your stream is encoded and sent via RTMP ingest. By the time your stream reaches the HLS distribution system, captions are already baked into the video frames as a visual overlay.
This means StreamTranslate captions work regardless of what delivery protocol the platform uses for viewers. Whether viewers receive your stream via HLS, DASH, WebRTC, or any future protocol, the captions are already part of the video and will be visible. There's no dependency on viewer clients supporting specific caption track formats or players having caption UI.
HLS does support embedded caption tracks via WebVTT, which enables true closed captions (toggleable by viewers). However, this requires either platform-native caption track generation or pushing caption data via the streaming platform's API — a more complex integration. StreamTranslate's visual overlay approach sacrifices the toggle feature in exchange for universal compatibility, simpler setup, and higher caption accuracy using Deepgram Nova-2.
StreamTranslate captions are baked into video before HLS encoding, ensuring they appear on every platform and every viewer device without compatibility concerns.
No changes to your RTMP or HLS settings are needed. StreamTranslate integrates via OBS browser source before the stream encoding step.
Because captions are in the video itself, they're served through CDN infrastructure with zero additional latency or infrastructure complexity.
HLS (HTTP Live Streaming) delivers video to viewers by breaking it into small segments over HTTP. It's used by Twitch, YouTube, and virtually all major streaming platforms.
RTMP pushes your stream from OBS to the server (ingest). HLS delivers from server to viewers (distribution). They handle different legs of the streaming pipeline.
Standard HLS adds 10-30 seconds of latency. Low-Latency HLS (LL-HLS) reduces this to 2-5 seconds on supported platforms and networks.
StreamTranslate generates caption overlays baked into your stream before HLS distribution, so captions are embedded in the video regardless of delivery protocol.
Yes, HLS supports WebVTT caption tracks. StreamTranslate's overlay approach instead bakes captions visually into video, which is simpler and works across all HLS players.