Video streaming glossary: 150+ terms — HLS, DASH, WebRTC, DRM

No terms match your filters. Try a different keyword or clear filters.

95th percentile billing

Bandwidth pricing model that bills based on the 95th percentile of 5-minute usage samples across a month, ignoring the top 5 % of spikes. The dominant billing model for transit and peering.

The HLS/DASH technique where the player picks a video rendition matching available bandwidth at each segment boundary, switching up or down as conditions change. The reason video doesn't freeze when your Wi-Fi sags.

ARQ (Automatic Repeat reQuest)

The loss-recovery strategy where the receiver detects missing packets and explicitly asks the sender to retransmit them. The mechanism behind TCP, SRT and RIST.

AVPlayer

Apple's native video-playback framework on iOS, macOS, tvOS, watchOS and visionOS. The only way to play HLS efficiently on Apple devices, with built-in FairPlay support.

BBR

Google's congestion-control algorithm that models bottleneck bandwidth and minimum RTT directly instead of reacting to loss. Higher throughput on lossy paths, lower latency under load.

Bitmovin Analytics

Bitmovin's QoE analytics product, tightly integrated with their player SDK but usable with hls.js, Shaka, Video.js, AVPlayer and ExoPlayer. Strong on detailed playback diagnostics.

Blocking playlist reload

LL-HLS server-side wait mechanism: the player asks for the next playlist version using `_HLS_msn=` and `_HLS_part=` query parameters; the server holds the response until that version exists.

BOLA

Buffer Occupancy based Lyapunov Algorithm — a buffer-based ABR with provable optimality bounds. Published by Spiteri et al. in 2016, default ABR in dash.js.

BrightScript

Roku's proprietary programming language for app development on Roku devices. The only way to build a Roku channel; its `Video` node handles HLS and DASH playback.

Buffer-based ABR

ABR algorithm that picks rendition based on current player buffer level, not measured throughput. Higher buffer = pick a higher rendition; lower buffer = pick a lower one.

Byte-range request

An HTTP request that asks for a specific byte range of a resource via the `Range:` header (RFC 9110, June 2022). Used in HLS to fetch sub-pieces of a CMAF file without naming each one as a separate URL.

Cache hit ratio (CHR)

The fraction of CDN requests served from cache without going to origin. The number every CDN dashboard tracks and every traffic engineer obsesses about.

Cache key

The set of request attributes (URL, headers, query string) that a CDN uses to identify a cache entry. Bad cache keys are the most common cause of low cache hit ratio.

CDM (Content Decryption Module)

The OS- or browser-provided component that performs DRM key handling and content decryption. The trusted execution boundary between encrypted media and the decoder.

CDN (Content Delivery Network)

A globally distributed network of edge servers that cache and deliver content close to viewers. The economic and performance foundation of every modern streaming service.

Chunked Transfer Encoding (CTE)

The HTTP/1.1 feature (RFC 9112, June 2022) that lets a server send a response body in pieces without knowing the total size in advance. The mechanism behind LL-HLS and LL-DASH low-latency delivery.

CMAF chunk

A small (typically 100–400 ms) addressable sub-piece of a CMAF segment. The atomic unit that LL-HLS partials and LL-DASH chunks both reference.

Congestion control

The algorithm a transport uses to decide how fast to send without overwhelming the network. CUBIC is the long-time TCP default; BBR is increasingly used for streaming.

Content steering

A standardised mechanism (HLS Content Steering, DASH Content Steering, both 2023) for a server to tell the player which CDN to fetch from, mid-session and dynamically.

Contribution

The first hop of the streaming pipeline — getting a live signal from the source (camera, switcher, OB van) to the encoder or origin. Optimised for reliability over one or two streams, not for fan-out to viewers.

Conviva

The longest-standing dedicated video QoE analytics company (founded 2006). Player-side SDK plus large-enterprise analytics dashboard. Default at Disney+, HBO Max, Paramount+ and most tier-1 OTT.

CSAI (Client-Side Ad Insertion)

Ad insertion done in the player: a separate ad player runs the ad stream between content segments. Easier to set up than SSAI but vulnerable to ad blockers.

CUBIC

The long-time default TCP congestion-control algorithm, standardised in RFC 8312 (February 2018). Loss-based: grows the congestion window along a cubic curve, halves on loss.

dash.js

The DASH-IF reference JavaScript player for MPEG-DASH. BSD 3-clause license, used as both production player and the canonical DASH conformance reference.

Distribution (streaming)

The fan-out leg of the pipeline — from origin or packager to millions of player endpoints via a CDN. Optimised for cache-friendly HTTP, low cost per GB, and broad reach across networks.

DTLS-SRTP

The encryption pair WebRTC uses for media: DTLS (RFC 9147, April 2022) negotiates keys, SRTP (RFC 3711, March 2004) encrypts RTP packets. End-to-end-encrypted between peers; mandatory in WebRTC.

Edge server

A CDN node located close to viewers — usually in a regional or city-level POP — that caches content and serves it to nearby clients. The last hop before the player.

EME (Encrypted Media Extensions)

W3C JavaScript API (Recommendation 2017) that lets a web video player negotiate licenses with a DRM CDM and play encrypted media. The browser-side counterpart of MSE for DRM-protected content.

End-to-end latency

The delay between the encoder output and the decoder input in a streaming chain. Narrower than glass-to-glass — excludes camera capture and display rendering.

Exits before video start

The percentage of sessions where the viewer abandoned before the first video frame rendered. The companion metric to startup time — measures the fraction of viewers who never see anything.

ExoPlayer

Google's open-source video player for Android, now distributed as part of AndroidX Media3. Plays HLS, DASH, SmoothStreaming, MP4 and many other formats with Widevine DRM.

EXT-X-DISCONTINUITY

The HLS playlist tag that tells the player the next segment has a discontinuity in timestamp, encoding parameters, or initialisation segment. Required at ad break boundaries and when switching content.

EXT-X-PART

The HLS playlist tag (HLS Authoring Spec rev 2020) that declares a partial segment — a 100–400 ms sub-piece of a full segment. The first-class tool for cutting LL-HLS latency.

FEC (Forward Error Correction)

A loss-recovery strategy where the sender adds redundant parity packets so the receiver can reconstruct losses without asking for retransmission. The latency-stable but bandwidth-expensive alternative to ARQ.

Forensic watermarking

Per-session imperceptible identifier embedded in the video bitstream that lets a content provider trace pirated copies back to the specific session that leaked them. Layered on top of DRM, not a replacement for it.

Geo-blocking

Restricting content access based on the viewer's geographic location, determined from their IP address. Required by most content-rights deals that grant per-territory rights.

Glass-to-glass latency

The full delay between light hitting the camera sensor (the source "glass") and that same image reaching the viewer's display (the destination "glass"). The industry's headline number for live latency.

HESP (High-Efficiency Streaming Protocol)

THEO Technologies' (now Dolby) HTTP-based low-latency protocol using a continuation stream plus initialisation stream. Targets sub-second latency over standard CDNs.

hls.js

The dominant open-source JavaScript HLS player. Apache 2.0 license, runs in any browser with MSE, used by Twitch, JW Player, Video.js and thousands of other deployments.

HTTP/2

The 2015 HTTP revision (RFC 9113, June 2022 consolidated) that brought binary framing, header compression and stream multiplexing over a single TCP connection. The transport that made low-latency HLS practical.

HTTP/3

HTTP semantics over QUIC instead of TCP. Eliminates HTTP/2's head-of-line blocking, supports 0-RTT and is now a standard option on major CDNs for streaming traffic.

ICE (Interactive Connectivity Establishment)

The framework (RFC 8445, July 2018) by which two WebRTC endpoints negotiate the best path between them, trying direct connections first and falling back to relays. The connectivity layer of WebRTC.

Ingest

The first-stage service that receives a contribution stream from a source and hands it to the encoder, packager or live origin. The on-ramp of a streaming pipeline.

Janus

A C-based general-purpose WebRTC gateway from Meetecho. Plugin-based architecture: SFU, streaming, recording and bridging are all plugins on top of a common core. GPL 3.0 license.

Jitsi Videobridge

The SFU at the core of the open-source Jitsi Meet conferencing platform. Java-based, Apache 2.0 license, deployed at scale by 8x8 (which acquired Jitsi in 2018).

Jitter

The variation in packet arrival times. A network with average 50 ms RTT but ±30 ms jitter delivers some packets at 20 ms and some at 80 ms, forcing receivers to buffer to smooth it out.

Just-in-time packaging

A workflow where CMAF segments are generated on demand from a single mezzanine asset rather than pre-baked into HLS and DASH formats. Saves storage at the cost of CPU at the edge or origin.

LiveKit

A Go-based WebRTC SFU with a higher-level API (rooms, participants, tracks, signalling baked in). Open-source (Apache 2.0) with a managed cloud product. ChatGPT Voice runs on it.