Video Streaming Glossary: Codecs, Protocols and Delivery A-Z

A video streaming glossary is a plain-English list of the terms you meet when you encode, package, deliver and play video online. This one defines 114 terms, from AAC to Widevine, and links each to its deep post on the AIOZ Stream knowledge hub where one is live.

TL;DR:

• Terms are sorted A to Z with one heading per letter, so you can jump with the letter index or your browser's find (Ctrl+F or Cmd+F).
• A linked term opens the full explainer on this hub; an unlinked term has no deep post yet, and its link will be added when one goes live.
• Definitions are vendor-neutral, and where a term touches AIOZ Stream, the entry states only what the AIOZ Stream docs say.
• Every definition was checked against the sources grouped under References at the bottom of the page.

How do you use this video streaming glossary?

You can read the page top to bottom, but most people arrive with one acronym in mind. Jump straight to a letter: A · B · C · D · E · F · G · H · I · L · M · O · P · Q · R · S · T · U · V · W. There are no entries yet for J, K, N, X, Y or Z; keyframe sits under I-frame.

Each entry starts with the term in bold, then gives a definition of one to three sentences in our own words. When the term is a link, the deep post goes further, with diagrams, examples and the trade-offs you need before you choose a setting.

The diagram below shows where the main terms sit in a streaming pipeline, so you can see which stage a word belongs to before you look it up.

Which streaming terms should you learn first?

If you are new to video, learn the terms in the order the video travels. Start with codec, container and bitrate, because every other choice depends on them. Then read bitrate ladder, ABR and rendition to see how one upload becomes several qualities.

Next come segment, manifest, HLS and DASH, which explain how players fetch video in small pieces over HTTP. Finish with CDN, edge node, latency and QoE, which cover how the video reaches your viewers and how you measure whether they enjoyed it.

A

AAC (Advanced Audio Coding) is the lossy audio codec from MPEG that nearly every browser and device can play, which makes it the safe default audio track for streaming video. The two profiles you will meet most are AAC-LC and HE-AAC. AIOZ Stream's audio encoding settings list "aac" as the supported codec.

ABR (adaptive bitrate) streaming publishes the same video at several bitrates and resolutions so the player can switch between them as your bandwidth changes. The idea is to drop a little picture quality before the viewer ever sees a buffering spinner.

AC-3 and E-AC-3 (Dolby Digital and Dolby Digital Plus) are Dolby's family of multichannel audio codecs, with E-AC-3 as the newer and more efficient member. You mostly meet them as surround sound tracks in TV apps and home theater devices.

Alternate audio means extra audio tracks for the same video, such as other languages or a commentary track, that your viewer can pick in the player. In HLS, these tracks are declared with the EXT-X-MEDIA tag in the multivariant playlist.

AOM (Alliance for Open Media) is the industry group, founded by companies including Google, Amazon, Netflix, Intel and Mozilla, that builds royalty-free media formats for the open web. Its best-known output is the AV1 video codec.

Aspect ratio is the proportion between a video's width and height, written as width:height. A 1920x1080 frame is 16:9 and vertical phone video is usually 9:16, and you can work it out by dividing both sides by their greatest common divisor.

AV1 is the royalty-free video codec from the Alliance for Open Media and the spiritual successor to Google's VP9. It holds up well at very low bitrates, but encoding it takes far more compute, so check device decode support before you rely on it.

AVC (H.264) is the video codec standard that ITU-T and MPEG developed together, and the two names mean the same thing. It still plays on almost every browser and device, and AIOZ Stream's video encoding settings accept it as "h264".

AVOD (advertising-based video on demand) is a model where you watch a catalog of on-demand content for free and the service earns money from the ads it shows. It sits beside SVOD and TVOD as one of the three main ways to sell VOD.

B

B-frame (bidirectional predicted frame) stores only the differences from frames both before and after it, which usually makes it the smallest frame type in a group of pictures. The cost is that the decoder needs those later frames first, which adds a little delay.

Beacon is a small analytics message that a video player sends to a data service, usually as an HTTP GET with query parameters or as a POST body. Beacons are how player events such as plays, errors and stalls reach your quality dashboard.

Bitrate is the amount of data used per second of video, usually given in kilobits or megabits per second. A higher bitrate generally means better quality and bigger files, and AIOZ Stream's encoding settings take your target bitrate in bits per second.

Bitrate ladder (encoding ladder) is the set of renditions you encode for adaptive streaming, each rung with its own resolution and bitrate. The player climbs up or down the rungs as the viewer's network conditions change.

Buffering is the wait you experience while the player downloads enough video to start or keep playing. It is widely seen as the biggest single reason viewers abandon a stream, which is why streaming teams track it so closely.

C

Cache is a stored copy of data kept close to where it is needed so the next request is faster. In video delivery, CDN edge servers cache segments near your viewers and browsers cache recent requests, so popular videos rarely go back to the origin.

CBR (constant bitrate) is a rate control mode that holds the bitrate steady for the whole stream, whatever the scene complexity. It is the usual choice for live ingest, because sudden bitrate swings can make the player fall behind and stall.

CDM (Content Decryption Module) is the browser or device component that holds DRM keys and decrypts protected media for playback. Your web page never handles those keys directly; it talks to the CDM through Encrypted Media Extensions.

CDN (content delivery network) is a network of servers in many locations that keeps copies of files close to viewers. When you press play, segments come from a nearby edge server instead of a distant origin, which cuts startup time and origin load.

CEA-608 and CEA-708 are the US broadcast closed caption standards. CEA-608 captions were carried on line 21 of analog NTSC video, while CEA-708 (now ANSI/CTA-708-E) carries captions in digital TV streams and includes 608 data for older receivers.

CENC (Common Encryption) is the MPEG standard for encrypting audio and video once inside ISO BMFF (MP4) files so that several DRM systems, such as Widevine and PlayReady, can unlock the same files. It saves you from storing a separate copy for each DRM system.

Chapters split a video into named sections that viewers can jump to, and on the web they are usually stored as a WebVTT file of timed cues. AIOZ Stream stores one chapter file per language for each media item, uploaded as a .vtt file.

Closed captions are a text version of a video's speech and important non-speech sounds, synchronized with the audio and switched on or off by the viewer. They serve people who are Deaf or hard of hearing, while subtitles mainly translate dialogue for people who do not speak the language.

CMAF (Common Media Application Format) is the ISO/IEC 23000-19 standard for packaging media as fragmented MP4, so one set of files can serve both HLS and DASH players. It also defines small chunks inside each segment, which is what makes low-latency CMAF delivery possible.

Codec (coder-decoder) is a method for compressing audio or video for storage and transmission, and for decompressing it again for playback. Most streaming codecs, such as H.264, HEVC and AV1, are lossy: they discard detail you are unlikely to notice in order to save bandwidth.

Container format is a file format that wraps one or more audio, video and subtitle streams together with metadata, so they can be stored and played as one file. MP4, WebM, MOV and MKV are containers, and the codecs inside them are a separate choice you make.

CRF (constant rate factor) is a quality-based rate control mode in encoders such as x264 and x265: you pick a quality level and the encoder spends whatever bitrate each scene needs. In x264 the scale runs from 0 to 51 with 23 as the default, and lower values mean higher quality.

D

DASH (MPEG-DASH), short for Dynamic Adaptive Streaming over HTTP, is the MPEG standard for adaptive streaming from ordinary web servers. It describes the available renditions in an XML manifest called an MPD, and AIOZ Stream's video API can output it with fMP4 segments.

Decentralized live streaming delivers a live broadcast through a distributed network of independently run nodes instead of one company's servers and CDN. AIOZ Stream's docs describe its live streaming as running on decentralized and scalable infrastructure.

Decoding is the reverse of encoding: the player turns compressed video data back into frames it can show on screen. Your device can decode in dedicated hardware or in software, and hardware support for a codec often decides whether playback stays smooth.

Demuxing and muxing are two halves of one job. Muxing (multiplexing) combines separate video, audio, caption and metadata streams into one container, such as an MPEG-TS segment or an MP4 file, and demuxing splits that container back into its streams.

DePIN (decentralized physical infrastructure network) is a network of physical hardware, such as storage, bandwidth and compute, run by many independent operators instead of one company. AIOZ Stream's docs describe AIOZ DePIN as a "Decentralized Content Delivery Network" for video storage and streaming.

DRM (digital rights management) protects video so that it plays only under the rules you set, such as a rental window or a minimum device security level. The three DRM systems you will meet most often are Google Widevine, Apple FairPlay and Microsoft PlayReady.

DRM license is what a player must fetch from a license server before it can play protected video. It sets the playback rules, such as an expiry time, geographic limits or a requirement for HDCP on the display link.

DVR (live rewind) lets you pause, rewind and seek within a live stream, often back to the start of the event. The name comes from the digital video recorders that offered the same controls for broadcast cable.

E

Edge node is the CDN server closest to the viewer. On the first request for a file it fetches a copy from the origin, serves it and keeps it, so the next viewers in that region get the file much faster.

Embed (video embed) is a snippet of code, usually an iframe, that places a player from a video service on your own web page. You copy the code from the platform and paste it into your HTML, and the video plays without your viewers leaving your site.

EME (Encrypted Media Extensions) is the W3C browser API that lets a web page play encrypted media by connecting the HTML video element to a Content Decryption Module. It is how DRM systems such as Widevine and PlayReady work inside the browser without plugins.

Encoding compresses raw video frames into a much smaller stream by following the rules of a codec. People often say encoding when they mean transcoding, which decodes a video that is already compressed and encodes it again.

Exit before video start (EBVS) is the share of viewers who press play and leave before the first frame appears. Play counts and watch time miss these viewers completely, so EBVS is often the clearest signal that your startup is too slow.

F

FairPlay Streaming (FPS) is Apple's DRM for HLS. It lets content owners encrypt content, exchange keys securely and protect playback on iPhone, iPad, Apple TV and Mac.

FFmpeg is the open source multimedia framework that can decode, encode, transcode, mux, demux, stream and filter almost any audio or video format. If you script your own encoding or packaging jobs, you will most likely do it with its command-line tool, ffmpeg.

fMP4 (fragmented MP4) splits an MP4 file into a short header plus a series of fragments, so a player can start on any fragment without the whole file. It is the segment format behind DASH and CMAF, and AIOZ Stream's video API uses it as the container type for DASH output.

Frame rate (fps) is how many still frames a video shows each second, written as frames per second. Common values are 24, 30 and 60 fps, and higher rates make motion look smoother at the cost of more bitrate.

G

Glass-to-glass latency is the total delay from a camera lens capturing an action to that action appearing on your viewer's screen. It adds up the time spent in encoding, ingest, transcoding, packaging, delivery and the player buffer.

GOP (group of pictures) is a run of frames that starts with an I-frame and continues with P-frames and B-frames that depend on it. GOP length sets how often a player can start decoding or switch renditions, so encoders usually align it with segment length.

H

H.265 (HEVC), short for High Efficiency Video Coding, is the successor to H.264 from MPEG and ITU-T, and it compresses noticeably better, especially above 1080p. AIOZ Stream's video encoding settings accept it as "h265".

H.266 (VVC), short for Versatile Video Coding, is the codec that ITU-T and MPEG finalized in July 2020 as the successor to HEVC. It promises better compression again, but browser and device support is still limited.

HDCP (High-bandwidth Digital Content Protection) encrypts the link between a playback device and a display, such as an HDMI cable. A DRM license can require it, so protected video is never sent to a screen that would not protect it.

HDR (high dynamic range) video carries a much wider range of brightness than standard dynamic range, so bright highlights and dark shadows keep their detail. Common delivery formats include HDR10, HLG and Dolby Vision, and your viewer needs an HDR-capable screen to see the difference.

HLS (HTTP Live Streaming) is Apple's adaptive streaming protocol, published as RFC 8216, which delivers video as short segments listed in M3U8 playlists over plain HTTP. It plays on almost every device, and "hls" is the default output type in AIOZ Stream's video API.

HLS encryption (AES-128 and SAMPLE-AES) protects HLS media with AES-128, either by encrypting whole segments or, with SAMPLE-AES, individual media samples, and points to the key in the playlist's EXT-X-KEY tag. It protects content in transit but is not DRM, because anyone who can fetch the key can decrypt the video.

HTML5 video element (<video>) is the browser's built-in media player tag, which plays a file or stream without plugins. For HLS and DASH in most browsers, a JavaScript player feeds it data through Media Source Extensions.

I

I-frame (keyframe, IDR frame) is an intra-coded frame: a complete picture that decodes without reference to any other frame. An IDR frame is a special I-frame that also tells the decoder to drop earlier references, which makes it a clean starting point for a new segment.

iframe (inline frame) is the HTML element that loads one web page inside another, and it is the most common way to embed a video player from another service. Watch the spelling: it has nothing to do with an I-frame in video compression.

Ingest is the step where video enters a streaming service, whether that is an uploaded file or a live feed pushed from an encoder. For live video, ingest usually means software such as OBS Studio sending RTMP to the platform's ingest URL.

L

Latency in streaming is the delay between something happening in front of the camera and your viewer seeing it. One common tiering treats more than 20 seconds as standard latency and roughly 5 to 12 seconds as low latency, with real-time protocols going well below that.

Live edge is the newest point in a live stream that is ready to play, set by the most recent complete segment in the playlist. Players usually sit a few segments behind it to avoid stalling, and that gap is part of your latency.

Live to VOD turns a finished livestream into an on-demand video that viewers can watch later. On AIOZ Stream, you turn on "Save Stream to Archive When Finished" so the recording is saved as a standard video after the broadcast ends.

LL-HLS (Low-Latency HLS) is Apple's low-latency extension of HLS, which publishes partial segments and updates playlists faster to cut delay to a few seconds. It keeps the scale of HTTP and CDN delivery that plain HLS offers.

llms.txt is a proposed standard: a Markdown file at a site's root that gives super intelligence tools a structured map of the site's most useful content. AIOZ Stream publishes one for its docs so coding assistants can find the right SDK methods, parameters and authentication patterns.

M

M3U8 (multivariant playlist) is the UTF-8 playlist file format that HLS uses. A multivariant playlist, once called the master playlist, lists every rendition, and each media playlist lists the segments for one rendition.

Manifest is the text file that tells a player which renditions and segments exist and where to fetch them. HLS uses M3U8 playlists and DASH uses an XML MPD, and your player reads the manifest before it downloads any video.

MCP (Model Context Protocol) is an open standard from Anthropic for connecting super intelligence applications to external tools and data through MCP servers. AIOZ Stream's docs describe an AIOZ MCP Server that MCP clients reach over SSE for media and live streaming operations.

Mesh, SFU and MCU are three ways to connect the people in a WebRTC session. In a mesh every peer sends to every other peer, an SFU (selective forwarding unit) forwards streams without decoding them, and an MCU (multipoint control unit) mixes them into one stream.

MoQ (Media over QUIC) is a media delivery protocol that the IETF moq working group is developing on top of QUIC. Its charter calls for a simple, low-latency solution for both ingest and distribution, covering live streaming, gaming and media conferencing, and the specifications are still drafts.

MPD (Media Presentation Description) is the XML manifest of an MPEG-DASH stream. It describes each segment's timing, URL and media characteristics such as resolution and bitrate, and it can address segments through a SegmentTemplate, SegmentList or SegmentTimeline.

MPEG-TS (MPEG transport stream) is a container built for broadcast that multiplexes audio and video into one stream of small packets. It is the classic HLS segment format, and "mpegts" is the default HLS container type in AIOZ Stream's video API.

MSE (Media Source Extensions) is the browser API that lets JavaScript build a media stream piece by piece and play it in an HTML audio or video element. JavaScript players use it to play HLS and DASH in browsers that do not support those formats natively.

Multi-CDN means delivering video through more than one CDN and routing each request to the one that performs best, costs least or is still up. It protects you from a single provider's outage or weak region.

Multicast (AIOZ Stream restreaming) in AIOZ Stream's API means sending one live broadcast to several RTMP destinations, such as YouTube, Facebook and Twitch, at the same time. You attach the destination URLs to a stream key, so you need only one encoder and one outgoing stream. Network engineers use the word for IP multicast, which is a different idea.

O

Opus is a royalty-free audio codec from the Xiph.Org Foundation that handles both speech and music with low delay. That low delay suits real-time communication, and in video it is most often paired with VP9 or AV1.

Origin server holds the original copies of your video files, and CDN edge servers pull from it when they do not have a cached copy. A high cache hit rate keeps load and egress traffic off your origin.

OVP (online video platform) is a hosted service that bundles video upload, encoding, storage, a player, delivery and analytics. Read literally it covers any video site, but in the industry it usually means business platforms you build on rather than consumer sites.

P

Per-title encoding builds a separate bitrate ladder for each video based on how complex it is, instead of one fixed ladder for the whole catalog. Netflix described the idea in 2015, and it either raises quality at the same file size or cuts size at the same quality.

PlayReady is Microsoft's content protection technology for distributing audio and video more securely and preventing unauthorized use. It is one of the three DRM systems most streaming services support, next to Widevine and FairPlay.

Program date time (EXT-X-PROGRAM-DATE-TIME) is the HLS tag that ties the first sample of a segment to an absolute date and time. Players and tools use it to sync streams, line up ads and map a live stream to the wall clock.

PSNR and SSIM are older objective quality metrics that compare an encoded frame with its source, PSNR (peak signal-to-noise ratio) pixel by pixel and SSIM (structural similarity) by structure. They are cheap to compute but follow human judgment less closely than VMAF.

Q

QoE (quality of experience) measures streaming from the viewer's side: how fast playback starts, how often it stalls, how sharp the picture looks and whether it fails. A short network outage that the buffer hides shows up in QoS numbers but not in QoE.

QoS (quality of service) measures how well the network and servers deliver bytes, through numbers such as bandwidth, throughput, latency and availability. It describes the delivery system, while QoE describes what your viewer felt.

QUIC is the transport protocol that Google first built and the IETF standardized, which runs over UDP yet gives reliable, encrypted and multiplexed streams. Packet loss on one stream does not stall the others, and QUIC is the transport under HTTP/3 and Media over QUIC.

R

Rate limit caps how many API requests you can make in a time window, and going over it returns HTTP 429 Too Many Requests. AIOZ Stream allows 100 upload, 200 write and 500 read requests per minute, and it reports your limits in X-RateLimit headers.

Rebuffering ratio is the share of a viewing session spent stalled after playback has started. AIOZ Stream's QoE docs define it as total stall time divided by watch time, and they note that even one two-second freeze can break immersion.

Rendition is one version of a video in an adaptive set, usually differing from the others in resolution and bitrate. Renditions can also differ by codec, frame rate or language, and some are audio-only.

Resolution is a video's width and height in pixels, such as 1280x720 (720p) or 3840x2160 (4K). AIOZ Stream's video API accepts output resolutions from 240p up to 4320p, and it ignores any quality set higher than the source.

RTMP (Real-Time Messaging Protocol) is the TCP-based protocol, born in Adobe Flash, that most encoders still use to push a live stream to a platform. You configure it as a server URL plus a stream key, which is how AIOZ Stream ingests live video.

RTMP pass-through forwards an incoming RTMP stream to one or more other RTMP destinations as it arrives, a feature also sold as simulcast or rebroadcast. It differs from live transcoding, where the platform re-encodes the stream into new renditions.

RTP (Real-time Transport Protocol) carries audio and video packets over UDP. It is a building block rather than a complete ingest protocol, and other protocols layer on top of it.

S

SDH (subtitles for the deaf and hard of hearing) are same-language subtitles that add speaker names, sound effects and music cues to the dialogue. They serve the same viewers as closed captions, but they are formatted and delivered like subtitles.

Segment (media segment) is a short piece of a video, typically a few seconds long, that the player downloads with one HTTP request. Each HLS media playlist lists its segments in order, with a duration for each in the EXTINF tag.

SRT (Secure Reliable Transport) is an open source, UDP-based contribution protocol from Haivision that recovers lost packets inside a latency buffer you can configure. Broadcasters use it as an alternative to RTMP when the network between encoder and platform is unreliable.

SRT (SubRip subtitle file) is a simple text subtitle format made of numbered cues, each with a start time, an end time and a line of text. It shares an acronym with the transport protocol above but has nothing to do with it, and on the web WebVTT is the native format.

SSAI and CSAI (server-side and client-side ad insertion) are the two ways to put ads into a stream. SSAI stitches ads into the video or manifest on the server so the viewer gets one continuous stream, while CSAI has the player request and play the ads itself.

Startup time is how long a viewer waits between pressing play and seeing the first frame. AIOZ Stream's QoE docs list it as a core metric and note that viewers feel the difference between 1.2 and 3 seconds.

Stream key is the secret credential your encoder sends with the RTMP server URL so the platform knows which channel the stream belongs to. On AIOZ Stream a key can be reused across broadcasts, and anyone who has it can stream to your channel, so keep it private.

SVOD (subscription video on demand) charges you a recurring fee for access to a whole catalog of on-demand content. It trades per-title revenue for steadier income from subscribers.

T

TCP and UDP are the two core internet transport protocols. TCP guarantees ordered, complete delivery by resending lost packets, while UDP sends without those guarantees and lets protocols such as SRT and QUIC add only the recovery they need.

Transcoding decodes a video that is already compressed and encodes it again, usually to create the renditions of an adaptive bitrate ladder or to change codec. When you upload one file and get several qualities back, transcoding is the step that made them.

Transmuxing (transmultiplexing) moves the same audio and video streams into a different container, such as MPEG-TS to MP4, without re-encoding them. Because nothing is decoded, it is fast and loses no quality.

TVOD (transactional video on demand) charges you per title, either to rent it for a limited window or to buy it outright. It suits new releases and one-off events where a subscription would not make sense.

U

UGC (user-generated content) is any content that the users of a platform create and upload, as opposed to content the platform licenses or produces itself. If you run a UGC video service, you have to plan for unpredictable formats, volumes and moderation.

V

VBR (variable bitrate) lets the encoder spend more bits on complex scenes and fewer on simple ones, so quality stays even across the video. It gives better quality per byte than CBR for on-demand video, but its bitrate peaks are harder to handle in live delivery.

Video API is a programming interface that lets your application upload, encode, store, deliver and measure video with code instead of a dashboard. AIOZ Stream's API covers media, players, playlists, webhooks, analytics and live streaming.

VMAF (Video Multi-Method Assessment Fusion) is Netflix's open source perceptual quality metric, which predicts how a person would rate an encoded video on a 0 to 100 scale. It won an Emmy, and AIOZ Stream's QoE docs use it as their example of perceived quality.

VOD (video on demand) is video that viewers can start whenever they choose, as opposed to live or scheduled broadcasts. On-demand libraries, course videos and replays of finished livestreams are all VOD.

VP8 and VP9 are Google's royalty-free video codecs, with VP9 as the successor to VP8 and a rival to HEVC. YouTube uses VP9 extensively, and it plays in Chrome, Firefox, Edge and most Android devices.

W

Watch time and retention tell you how much of your content people watch and how long they stay. In AIOZ Stream's Analytics API, watch_time is the cumulative seconds a user spent watching and retention is the total time a user stayed continuously engaged.

Watch-to-Earn (W2E) is a monetization model in which viewers earn rewards for the time they spend watching. The linked explainer sets out what AIOZ has and has not published about W2E on AIOZ Stream.

WCAG (Web Content Accessibility Guidelines) is the W3C standard for accessible web content. For video, it sets requirements for captions, audio description and transcripts at conformance levels A, AA and AAA.

Webhook is an HTTP callback: when an event happens, the service sends a POST request with event data to a URL you registered. AIOZ Stream uses webhooks so your application hears about events as they occur instead of polling the API.

WebM is a royalty-free container format, a subset of Matroska, built for VP8, VP9 and AV1 video with Opus or Vorbis audio. You see it most often in web video, where browsers can play it natively.

WebRTC (Web Real-Time Communication) is a set of W3C and IETF standards that lets browsers and apps send audio, video and data between peers in real time. It powers video calls and the lowest-latency live streams, but it is harder to scale than HTTP streaming.

WebVTT (Web Video Text Tracks) is the W3C text format for timed cues such as captions, subtitles and chapters, and browsers can display it in the video element. AIOZ Stream accepts .vtt files for media transcripts and chapters.

WHIP and WHEP are simple HTTP-based protocols for setting up WebRTC sessions: WHIP (WebRTC-HTTP Ingestion Protocol) for publishing a stream and WHEP (WebRTC-HTTP Egress Protocol) for watching one. They let WebRTC work with encoders, streaming services and CDNs in a standard way.

Widevine is Google's DRM system for protecting premium video in browsers, phones and TV devices. It is usually paired with CENC-encrypted media, so the same files can also work with PlayReady.

How AIOZ Stream handles these terms

Many of the terms above map directly onto fields in the AIOZ Stream API. When you create a video object, each quality you request names a resolution, a streaming type (hls or dash), a container type and optional video and audio settings. The docs list h264 and h265 as video codecs, aac as the audio codec, and mpegts or mp4 containers for HLS with fmp4 for DASH.

This is the quality object from the create-video request body in the docs, with the placeholder values replaced by values the same page lists as supported:

{
  "resolution": "1080p",
  "type": "hls",
  "container_type": "mpegts",
  "video_config": { "codec": "h264", "bitrate": 3000000, "index": 0 },
  "audio_config": { "codec": "aac", "bitrate": 128000, "channels": "2", "sample_rate": 48000, "index": 0 }
}

For live video, you send RTMP from an encoder with a stream key, and you can multicast the same broadcast to other RTMP destinations. Delivery runs on AIOZ DePIN, which the docs call a Decentralized Content Delivery Network. To try these fields on your own uploads, start with the AIOZ Stream Video API reference.

Video streaming glossary FAQ

What is the difference between a codec and a container?
A codec compresses the audio or video itself, while a container is the file that holds the compressed streams together. H.264 is a codec and MP4 is a container, so you can have H.264 video inside MP4, MPEG-TS or MKV files.

What is the difference between HLS and DASH?
Both are adaptive streaming formats that deliver short segments over HTTP. HLS comes from Apple and uses M3U8 playlists, while DASH is an MPEG standard that uses an XML manifest called an MPD.

What does ABR mean in streaming?
ABR means adaptive bitrate: the video is encoded at several bitrates and the player switches between them as your network speed changes. It trades a little sharpness for fewer stalls.

Is latency the same as buffering?
No. Latency is how far behind real time the viewer is, while buffering is a pause in playback while the player waits for data; a stream can have high latency and never buffer.

What is the difference between captions and subtitles?
Captions transcribe the speech and important sounds for viewers who cannot hear the audio. Subtitles mainly translate dialogue for viewers who can hear it but do not understand the language.

Related Reading

• What Is Adaptive Bitrate (ABR) Streaming? How ABR Works: the starting point for adaptive bitrate, quality and QoE
• Video Transcoding Basics: H.264 vs H.265 (HEVC) vs VP9: the starting point for codecs and transcoding
• HLS vs WebRTC: Which Streaming Protocol Should You Choose?: the starting point for streaming protocols and latency
• AIOZ Stream Video Player: Features and Customization Guide: the starting point for players and playback
• WCAG Video Requirements for Streaming and Video Platforms: the starting point for captions and accessibility
• Live Streaming on AIOZ Stream: The Complete Setup Guide: the starting point for live streaming
• What Is a CDN Edge Node? How Edge Caching Works for Video: the starting point for CDN and delivery
• What Is Video on Demand (VOD)? How VOD Streaming Works: the starting point for VOD and business models
• AIOZ Stream Developer Guide: API, SDKs, and Webhooks: the starting point for the AIOZ Stream developer platform
• AIOZ Stream Explained: Decentralized Video Infrastructure: the starting point for the whole platform, from upload to playback

References

AIOZ Stream documentation

• AIOZ Stream Docs: Audio API
• AIOZ Stream Docs: Video API
• AIOZ Stream Docs: Media Chapter API
• AIOZ Stream Docs: Live Streaming concepts
• AIOZ Stream Docs: AIOZ DePIN benefits for video storage and streaming
• AIOZ Stream Docs: Create a Livestream
• AIOZ Stream Docs: Live Streamings Media API
• AIOZ Stream Docs: LLM Docs (llms.txt)
• AIOZ Stream Docs: AIOZ MCP Server
• AIOZ Stream Docs: Livestream Multicast
• AIOZ Stream Docs: QoE Metrics
• AIOZ Stream Docs: Rate limits
• AIOZ Stream Docs: Analytics API
• AIOZ Stream Docs: Webhooks
• AIOZ Stream Docs: Media Transcript API

Standards and specifications

• IETF RFC 8216: HTTP Live Streaming
• ISO/IEC 23000-19:2024 (CMAF)
• W3C: Encrypted Media Extensions
• W3C WAI: Captions/Subtitles
• W3C WAI: Making Audio and Video Media Accessible
• IETF: Media Over QUIC (moq) working group charter
• IETF draft: WebRTC-HTTP Egress Protocol (WHEP)
• webrtc.org

Technology owners

• Apple Developer: FairPlay Streaming
• Apple Developer: About CMAF with HLS
• Microsoft Learn: PlayReady
• Google Widevine
• Netflix VMAF on GitHub
• Netflix Partner Help: subtitles, SDH and CC
• FFmpeg: About
• FFmpeg Wiki: H.264 encoding guide (CRF)
• llmstxt.org: the /llms.txt proposal

Reference guides

• MDN: the video element
• MDN: Media container formats
• MDN: Media Source Extensions API
• MDN: WebVTT API
• Wikipedia: Group of pictures
• Wikipedia: Dynamic Adaptive Streaming over HTTP
• IAB Digital Video Glossary (PDF)
• FADGI: Audio-Visual Accessibility guidelines (PDF)
• Stripe: OTT business models

Mux Video Glossary entries

• AAC, ABR, AC-3 and E-AC-3, Alternate audio, AOM, Aspect ratio, AV1, AVC, Beacon, Bitrate, Bitrate ladder, Buffering, Cache, CBR, CDN, CENC, Codec, DASH, Decoding, Demuxing and muxing, DRM, DRM license, DVR, Edge node, Embed, Encoding, Frame rate, Glass-to-glass latency, GOP, H.265, H.266, HDCP, HDR, HLS encryption, iframe, Ingest, Live edge, LL-HLS, M3U8, Manifest, MCP, MPEG-TS, Multi-CDN, Opus, Origin server, OVP, Per-title encoding, QoE, QoS, QUIC, Rebuffering ratio, Rendition, Resolution, RTMP, RTMP pass-through, RTP, SRT, SSAI and CSAI, TCP and UDP, Transcoding, Transmuxing, UGC, VBR, Video API, VOD, VP8 and VP9, Webhook, WebM, WebRTC