Dynamic packaging in Media Services v3

Azure Media Services provides built-in origin server and packaging capabilities to deliver content in HLS and MPEG DASH streaming protocol formats. In AMS, the streaming endpoint acts as the "origin" server sending formatted HLS and DASH content to client players that support adaptive bitrate streaming using those popular formats. The Streaming Endpoint also supports many features such as just-in-time, dynamic packaging with or without content protection, to reach all major devices (like iOS and Android devices).

Most browsers and mobile devices on the market today support and understand the HLS or DASH streaming protocols. For example, iOS requires streams to be delivered in HTTP Live Streaming (HLS) format and Android devices support HLS as well as MPEG DASH on certain models (or through the use of the application level player Exoplayer for Android devices.

In Media Services, a streaming endpoint (origin) represents a dynamic (just-in-time) packaging and origin service that can deliver your live and on-demand content directly to a client player app. It uses one of the common streaming media protocols mentioned in the following section. Dynamic packaging is a feature that comes standard on all streaming endpoints.

The advantages of just-in-time packaging are the following:

• You can store all your files in standard MP4 file format.
• You don't need to store multiple copies of static packaged HLS and DASH formats in blob storage which reduces the amount of video content stored and lowers your storage costs.
• You can instantly take advantage of new protocol updates and changes to the specifications without needing to re-package the static content in your catalog.
• You can deliver content with or without encryption and DRM using the same MP4 files in storage.
• You can dynamically filter or alter the manifests with simple asset-level or global filters to remove specific tracks, resolutions, languages, or provide shorter highlight clips from the same MP4 files without re-encoding or re-rendering the content.

To prepare your source files for delivery

To take advantage of dynamic packaging, encode your mezzanine (source) file into a set of single or multiple bitrate MP4 (ISO Base Media 14496-12) files. The encoded video files and streaming configuration files will be located in an output asset. From this set of files, you can use dynamic packaging to deliver video via the streaming media protocols.

Typically, you will use the Azure Media Services standard encoder to generate this content using the Content Aware Encoding presets, or the Adaptive Bitrate presets. Both generate a set of MP4 files ready for streaming and dynamic packaging.

Alternatively, you can choose to encode using an external service, on-premises, or on your own VMs or serverless function apps. Content encoded externally can be uploaded into an asset for streaming provided that it meets the encoding requirements for adaptive bitrate streaming formats. An example project of uploading a pre-encoded MP4 for streaming is available in the .NET SDK samples - see Stream Existing Mp4 files.

Azure Media Services dynamic packaging only supports video and audio files in the MP4 container format. Audio files must be encoded into an MP4 container as well when using alternate codecs like Dolby.

Make videos available for streaming

To make videos in the encoded asset available to clients for playback, publish the asset using a Streaming Locator and build the appropriate HLS and DASH streaming URLs. By changing the protocol used on the URL format query, the service will deliver the appropriate streaming manifest (HLS, MPEG DASH.)

As a result, you only need to store and pay for the files in single storage format (MP4) and Media Services will generate and serve the appropriate HLS or DASH manifests based on requests from your client players.

If you plan to protect your content by using Media Services dynamic encryption, see Streaming protocols and encryption types.

Deliver HLS

HLS dynamic packaging

Your streaming client can specify the following HLS formats. We recommend using the CMAF format for compatibility with the latest players and iOS devices. For legacy devices, the v4 and v3 formats are available as well by simply changing the format query string.

HLS packing ratio for VOD

To control the packing ratio of VOD content for older HLS formats, you can set the fragmentsPerHLSSegment metadata tag in the .ism file to control the default 3:1 packing ratio for TS segments delivered from the older v3 and v4 HLS format manifests. This setting change requires you to directly modify the .ism file in storage to adjust the packing ratio.

Example .ism server manifest with fragmentsPerHLSSegment set to 1.

   <?xml version="1.0" encoding="utf-8" standalone="yes"?>
   <smil xmlns="http://www.w3.org/2001/SMIL20/Language">
      <head>
         <meta name="formats" content="mp4" />
         <meta name="fragmentsPerHLSSegment" content="1"/>
      </head>
      <body>
         <switch>
         ...
         </switch>
      </body>
   </smil>

Deliver DASH

DASH dynamic packaging

Your streaming client can specify the following MPEG-DASH formats:

Deliver Smooth Streaming manifests

Smooth Streaming dynamic packaging

Your streaming client can specify the following Smooth Streaming formats:

On-demand streaming workflow

The following steps show a common Media Services streaming workflow where dynamic packaging is used along with the Standard Encoder in Azure Media Services.

1. Upload an input file such as a MP4, QuickTime/MOV, or other supported file format. This file is also referred to as the mezzanine or source file. For the list of supported formats, see Formats Supported by the Standard Encoder.

2. Encode your mezzanine file into an H.264/AAC MP4 adaptive bitrate set.

   If you already have encoded files and just want to copy and stream the files, use: CopyVideo and CopyAudio APIs. A new MP4 file with a streaming manifest (.ism file) will be created as a result.

   In addition, you can just generate the .ism and .ismc file on a pre-encoded file, as long as it is encoded using the right settings for adaptive bitrate streaming (this is typically 2-second GOPs, Key frame distances of 2s min and max, and Constant Bitrate (CBR) mode encoding.)

   See the stream existing Mp4 .NET SDK sample for details on how to generate the .ism (server manifest) and .ismc (client manifests) for streaming from an existing, pre-encoded MP4 file.

3. Publish the output asset that contains the adaptive bitrate MP4 set. You publish by creating a streaming locator.

4. Build URLs that target different formats (HLS, MPEG-DASH, and Smooth Streaming). The streaming endpoint would take care of serving the correct manifest and requests for all these different formats.

The download path is present in the above image just to show you that you can download an MP4 file directly through the streaming endpoint (origin) (you specify the downloadable streaming policy on the streaming locator).
The dynamic packager is not altering the file. You can optionally use the Azure blob storage APIs to access an MP4 directly for progressive downloading if you wish to bypass the streaming endpoint (origin) features.

Encode to adaptive bitrate MP4s

The following articles show examples of how to encode a video with Media Services:

• Use content aware encoding.
• Code samples for encoding with Standard Encoder using .NET

See the list of supported Standard Encoder input formats and codecs.

Live streaming workflow

A live event can be set to either a pass-through (an on-premises live encoder sends a multiple bitrate stream) or live encoding (an on-premises live encoder sends a single bitrate stream).

Here's a common workflow for live streaming with dynamic packaging:

1. Create a live event.
2. Get the ingest URL and configure your on-premises encoder to use the URL to send the contribution feed.
3. Get the preview URL and use it to verify that the input from the encoder is being received.
4. Create a new asset.
5. Create a live output and use the asset name that you created.
   The live output archives the stream into the asset.
6. Create a streaming locator with the built-in streaming policy types.
   If you intend to encrypt your content, review Content protection overview.
7. List the paths on the streaming locator to get the URLs to use.
8. Get the host name for the streaming endpoint you want to stream from.
9. Build URLs that target different formats (HLS, MPEG-DASH, and Smooth Streaming). The streaming endpoint takes care of serving the correct manifest and requests for the different formats.

For information about live streaming in Media Services v3, see Live streaming overview.

Video codecs supported by Dynamic Packaging

Dynamic packaging supports video files that are in the MP4 container file format and contain video that is encoded with H.264 (MPEG-4 AVC or AVC1) or H.265 (HEVC, hev1, or hvc1).

Audio codecs supported by dynamic packaging

Dynamic packaging also supports audio files that are stored in the MP4 file container format containing encoded audio stream in one of the following codecs:

• AAC (AAC-LC, HE-AAC v1, or HE-AAC v2).

• Dolby Digital Plus (Enhanced AC-3 or E-AC3). The encoded audio must be stored in the MP4 container format to work with Dynamic Packaging.

• Dolby Atmos

  Streaming Dolby Atmos content is supported for standards like the MPEG-DASH protocol with either Common Streaming Format (CSF) or Common Media Application Format (CMAF) fragmented MP4, and via HTTP Live Streaming (HLS) with CMAF.

• DTS
  DTS codecs supported by DASH-CSF, DASH-CMAF, HLS-M2TS, and HLS-CMAF packaging formats are:

  • DTS Digital Surround (dtsc)
  • DTS-HD High Resolution and DTS-HD Master Audio (dtsh)
  • DTS Express (dtse)
  • DTS-HD Lossless (no core) (dtsl)

Dynamic packaging supports multiple audio tracks with DASH or HLS (version 4 or later) for streaming assets that have multiple audio tracks with multiple codecs and languages.

For all of the above audio codecs, the encoded audio must be stored in the MP4 container format to work with Dynamic Packaging. The service does not support raw elementary stream file formats on blob storage (for example the following would not be supported - .dts, .ac3.)

Only files with the .mp4 of .mp4a extension are supported for audio packaging.

Limitations

iOS limitation on AAC 5.1 audio

Apple iOS devices do not support 5.1 AAC audio codec. Multi-channel audio must be encoded using Dolby Digital or Dolby Digital Plus codecs.

For detailed information, see HLS authoring specification for apple devices.

Dolby Digital audio

Media Services dynamic packaging does not currently support files that contain Dolby Digital (AC3) audio (as this is considered a legacy codec by Dolby).

Manifests

In Media Services dynamic packaging, the streaming client manifests for HLS, MPEG-DASH, and Smooth Streaming are dynamically generated based on the format query in the URL.

A manifest file includes streaming metadata such as track type (audio, video, or text), track name, start and end time, bitrate (qualities), track languages, presentation window (sliding window of fixed duration), and video codec (FourCC). It also instructs the player to retrieve the next fragment by providing information about the next playable video fragments that are available and their location. Fragments (or segments) are the actual "chunks" of video content.

Examples

HLS

Here's an example of an HLS manifest file, also called an HLS master playlist:

#EXTM3U
#EXT-X-VERSION:4
#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio",NAME="aac_eng_2_128041_2_1",LANGUAGE="eng",DEFAULT=YES,AUTOSELECT=YES,URI="QualityLevels(128041)/Manifest(aac_eng_2_128041_2_1,format=m3u8-aapl)"
#EXT-X-STREAM-INF:BANDWIDTH=536608,RESOLUTION=320x180,CODECS="avc1.64000d,mp4a.40.2",AUDIO="audio"
QualityLevels(381048)/Manifest(video,format=m3u8-aapl)
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=536608,RESOLUTION=320x180,CODECS="avc1.64000d",URI="QualityLevels(381048)/Manifest(video,format=m3u8-aapl,type=keyframes)"
#EXT-X-STREAM-INF:BANDWIDTH=884544,RESOLUTION=480x270,CODECS="avc1.640015,mp4a.40.2",AUDIO="audio"
QualityLevels(721495)/Manifest(video,format=m3u8-aapl)
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=884544,RESOLUTION=480x270,CODECS="avc1.640015",URI="QualityLevels(721495)/Manifest(video,format=m3u8-aapl,type=keyframes)"
#EXT-X-STREAM-INF:BANDWIDTH=1327398,RESOLUTION=640x360,CODECS="avc1.64001e,mp4a.40.2",AUDIO="audio"
QualityLevels(1154816)/Manifest(video,format=m3u8-aapl)
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=1327398,RESOLUTION=640x360,CODECS="avc1.64001e",URI="QualityLevels(1154816)/Manifest(video,format=m3u8-aapl,type=keyframes)"
#EXT-X-STREAM-INF:BANDWIDTH=2413312,RESOLUTION=960x540,CODECS="avc1.64001f,mp4a.40.2",AUDIO="audio"
QualityLevels(2217354)/Manifest(video,format=m3u8-aapl)
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=2413312,RESOLUTION=960x540,CODECS="avc1.64001f",URI="QualityLevels(2217354)/Manifest(video,format=m3u8-aapl,type=keyframes)"
#EXT-X-STREAM-INF:BANDWIDTH=3805760,RESOLUTION=1280x720,CODECS="avc1.640020,mp4a.40.2",AUDIO="audio"
QualityLevels(3579827)/Manifest(video,format=m3u8-aapl)
#EXT-X-I-FRAME-STREAM-INF:BANDWIDTH=3805760,RESOLUTION=1280x720,CODECS="avc1.640020",URI="QualityLevels(3579827)/Manifest(video,format=m3u8-aapl,type=keyframes)"
#EXT-X-STREAM-INF:BANDWIDTH=139017,CODECS="mp4a.40.2",AUDIO="audio"
QualityLevels(128041)/Manifest(aac_eng_2_128041_2_1,format=m3u8-aapl)

MPEG-DASH

Here's an example of an MPEG-DASH manifest file, also called an MPEG-DASH Media Presentation Description (MPD):

<?xml version="1.0" encoding="UTF-8"?>
<MPD xmlns="urn:mpeg:dash:schema:mpd:2011" xmlns:xsi="https://www.w3.org/2001/XMLSchema-instance" profiles="urn:mpeg:dash:profile:isoff-live:2011" type="static" mediaPresentationDuration="PT1M10.315S" minBufferTime="PT7S">
   <Period>
      <AdaptationSet id="1" group="5" profiles="ccff" bitstreamSwitching="false" segmentAlignment="true" contentType="audio" mimeType="audio/mp4" codecs="mp4a.40.2" lang="en">
         <SegmentTemplate timescale="10000000" media="QualityLevels($Bandwidth$)/Fragments(aac_eng_2_128041_2_1=$Time$,format=mpd-time-csf)" initialization="QualityLevels($Bandwidth$)/Fragments(aac_eng_2_128041_2_1=i,format=mpd-time-csf)">
            <SegmentTimeline>
               <S d="60160000" r="10" />
               <S d="41386666" />
            </SegmentTimeline>
         </SegmentTemplate>
         <Representation id="5_A_aac_eng_2_128041_2_1_1" bandwidth="128041" audioSamplingRate="48000" />
      </AdaptationSet>
      <AdaptationSet id="2" group="1" profiles="ccff" bitstreamSwitching="false" segmentAlignment="true" contentType="video" mimeType="video/mp4" codecs="avc1.640020" maxWidth="1280" maxHeight="720" startWithSAP="1">
         <SegmentTemplate timescale="10000000" media="QualityLevels($Bandwidth$)/Fragments(video=$Time$,format=mpd-time-csf)" initialization="QualityLevels($Bandwidth$)/Fragments(video=i,format=mpd-time-csf)">
            <SegmentTimeline>
               <S d="60060000" r="10" />
               <S d="42375666" />
            </SegmentTimeline>
         </SegmentTemplate>
         <Representation id="1_V_video_1" bandwidth="3579827" width="1280" height="720" />
         <Representation id="1_V_video_2" bandwidth="2217354" codecs="avc1.64001F" width="960" height="540" />
         <Representation id="1_V_video_3" bandwidth="1154816" codecs="avc1.64001E" width="640" height="360" />
         <Representation id="1_V_video_4" bandwidth="721495" codecs="avc1.640015" width="480" height="270" />
         <Representation id="1_V_video_5" bandwidth="381048" codecs="avc1.64000D" width="320" height="180" />
      </AdaptationSet>
   </Period>
</MPD>

Smooth Streaming

Here's an example of a Smooth Streaming manifest file:

<?xml version="1.0" encoding="UTF-8"?>
<SmoothStreamingMedia MajorVersion="2" MinorVersion="2" Duration="703146666" TimeScale="10000000">
   <StreamIndex Chunks="12" Type="audio" Url="QualityLevels({bitrate})/Fragments(aac_eng_2_128041_2_1={start time})" QualityLevels="1" Language="eng" Name="aac_eng_2_128041_2_1">
      <QualityLevel AudioTag="255" Index="0" BitsPerSample="16" Bitrate="128041" FourCC="AACL" CodecPrivateData="1190" Channels="2" PacketSize="4" SamplingRate="48000" />
      <c t="0" d="60160000" r="11" />
      <c d="41386666" />
   </StreamIndex>
   <StreamIndex Chunks="12" Type="video" Url="QualityLevels({bitrate})/Fragments(video={start time})" QualityLevels="5">
      <QualityLevel Index="0" Bitrate="3579827" FourCC="H264" MaxWidth="1280" MaxHeight="720" CodecPrivateData="0000000167640020ACD9405005BB011000003E90000EA600F18319600000000168EBECB22C" />
      <QualityLevel Index="1" Bitrate="2217354" FourCC="H264" MaxWidth="960" MaxHeight="540" CodecPrivateData="000000016764001FACD940F0117EF01100000303E90000EA600F1831960000000168EBECB22C" />
      <QualityLevel Index="2" Bitrate="1154816" FourCC="H264" MaxWidth="640" MaxHeight="360" CodecPrivateData="000000016764001EACD940A02FF9701100000303E90000EA600F162D960000000168EBECB22C" />
      <QualityLevel Index="3" Bitrate="721495" FourCC="H264" MaxWidth="480" MaxHeight="270" CodecPrivateData="0000000167640015ACD941E08FEB011000003E90000EA600F162D9600000000168EBECB22C" />
      <QualityLevel Index="4" Bitrate="381048" FourCC="H264" MaxWidth="320" MaxHeight="180" CodecPrivateData="000000016764000DACD941419F9F011000003E90000EA600F14299600000000168EBECB22C" />
      <c t="0" d="60060000" r="11" />
      <c d="42375666" />
   </StreamIndex>
</SmoothStreamingMedia>

Naming of tracks in the manifest

If an audio track name is specified in the .ism file, Media Services adds a Label element within an AdaptationSet to specify the textural information for the specific audio track. An example of the output DASH manifest:

<AdaptationSet codecs="mp4a.40.2" contentType="audio" lang="en" mimeType="audio/mp4" subsegmentAlignment="true" subsegmentStartsWithSAP="1">
  <Label>audio_track_name</Label>
  <Role schemeIdUri="urn:mpeg:dash:role:2011" value="main"/>
  <Representation audioSamplingRate="48000" bandwidth="131152" id="German_Forest_Short_Poem_english-en-68s-2-lc-128000bps_seg">
    <BaseURL>German_Forest_Short_Poem_english-en-68s-2-lc-128000bps_seg.mp4</BaseURL>
  </Representation>
</AdaptationSet>

The player can use the Label element to display on its UI.

Signaling audio description tracks

You can add a narration track to your video to help visually impaired clients follow the video recording by listening to the narration. You need to annotate an audio track as audio description in the manifest. To do that, add “accessibility” and “role” parameters to the .ism file. It's your responsibility to set these parameters correctly to signal an audio track as audio description. For example, add <param name="accessibility" value="description" /> and <param name="role" value="alternate" to the .ism file for a specific audio track.s

Smooth Streaming manifest

If you're playing a Smooth Streaming stream, the manifest would carry values in Accessibility and Role attributes for that audio track. For example, Role="alternate" Accessibility="description" would be added in the StreamIndex element to indicate it's an audio description.

DASH manifest

For DASH manifest, the following two elements would be added to signal the audio description:

<Accessibility schemeIdUri="urn:mpeg:dash:role:2011" value="description"/>
<Role schemeIdUri="urn:mpeg:dash:role:2011" value="alternate"/>

HLS playlist

For HLS v7 and above (format=m3u8-cmaf), its playlist would carry AUTOSELECT=YES,CHARACTERISTICS="public.accessibility.describes-video" when the audio description track is signaled.

Dynamic Manifest filtering

To control the number of tracks, formats, bitrates, and presentation time windows that are sent to players, you can use dynamic filtering with the Media Services dynamic packager. For more information, see Pre-filtering manifests with the dynamic packager.

Dynamic encryption for DRM

You can use dynamic encryption to dynamically encrypt your live or on-demand content with AES-128 or any of the three major digital rights management (DRM) systems: Microsoft PlayReady, Google Widevine, and Apple FairPlay. Media Services also provides a service for delivering AES keys and DRM licenses to authorized clients. For more information, see dynamic encryption.

Get help and support

You can contact Media Services with questions or follow our updates by one of the following methods:

• Q & A
• Stack Overflow. Tag questions with azure-media-services.
• @MSFTAzureMedia or use @AzureSupport to request support.
• Open a support ticket through the Azure portal.