Smooth Streaming client manifest elements

The client manifest is defined as the Manifest Response of Microsoft’s Smooth Streaming Protocol specification, MS-SSTR (revision 11.0, 2024). It is an XML 1.0 document with no DTD and no namespaces. Attributes may come in any order, and elements and attributes the specification does not name are allowed as vendor extensions. All numbers are decimal text.

SmoothStreamingMedia

The root element. Its content is an optional Protection element, then one or more StreamIndex elements.

Attribute	Type	Meaning
`MajorVersion`	required, `2`	Major version of the format.
`MinorVersion`	required, `0` or `2`	Minor version. Version 2.2 is needed for the `r` (repeat) attribute of fragments.
`TimeScale`	64-bit, default 10000000	Units per second of `Duration`: by default, 100 nanoseconds.
`Duration`	64-bit, required	Length of the presentation, in `TimeScale` units.
`IsLive`	`TRUE` or `FALSE`	A live presentation; otherwise on demand.
`LookaheadCount`	32-bit	Live only: the size of the server’s buffer, in fragments. Fragments carry the times of up to this many later fragments in their `tfrf` box.
`DVRWindowLength`	64-bit	Live only: how far back a viewer can seek, in `TimeScale` units; absent or 0 means without limit.

Protection

Protection holds one or more ProtectionHeader elements, one per content protection system.

Part	Meaning
`SystemID` attribute	The protection system’s UUID, such as `9A04F079-9840-4286-AB92-E65BE0885F95` for PlayReady. The specification’s grammar writes it in braces.
Content	Opaque data for that system, in Base64, such as a PlayReady header.

StreamIndex

One stream: the video, an audio track, or a text track. It holds one or more QualityLevel elements, then the c fragment elements described on Fragments and requests.

Attribute	Type	Meaning
`Type`	required: `video`, `audio`, `text`	The kind of stream.
`Subtype`	four letters; required for text	The use of a text stream (below).
`TimeScale`	64-bit	Units per second for this stream’s fragment times.
`Name`	letter, then letters, digits, `_` and `-`	The stream’s name, used in fragment addresses.
`Chunks`	32-bit	The number of fragments.
`QualityLevels`	32-bit	The number of tracks (bit rates).
`Url`	URL pattern	How to build fragment addresses, such as `QualityLevels({bitrate})/Fragments(video={start time})`.
`MaxWidth`, `MaxHeight`	32-bit; video only	Largest picture size in the stream.
`DisplayWidth`, `DisplayHeight`	32-bit; video only	Suggested display size.
`ParentStreamIndex`	a stream name	Makes this a sparse stream, timed by the named stream.
`ManifestOutput`	`TRUE` or `FALSE`	The stream’s sample data is written in the manifest itself.

Chunks, QualityLevels and Url are required except for an embedded track.

Subtype	Text stream holds
`SCMD`	Triggers for actions by the player application
`CHAP`	Chapter markers
`SUBT`	Subtitles for foreign-language audio
`CAPT`	Closed captions
`DESC`	Media descriptions
`CTRL`	Events that control the application
`DATA`	Other application data

QualityLevel

One track of the stream, at one bit rate. Its only possible child is a CustomAttributes element.

Attribute	Required for	Meaning
`Index`	all	Unique number of the track in the stream, normally from 0.
`Bitrate`	all	Average bit rate in bits per second; used as `{bitrate}` in addresses.
`MaxWidth`, `MaxHeight`	video, audio	Picture size in pixels.
`CodecPrivateData`	video, audio	Decoder set-up data in hexadecimal (below).
`FourCC`	audio	The codec (below).
`SamplingRate`, `Channels`, `BitsPerSample`	audio	As in the MP4 sample description.
`PacketSize`	audio	Size of each audio packet in bytes.
`AudioTag`	audio	A Windows wave format number (below).
`NALUnitLengthField`	H.264 only	Bytes in each NAL unit’s length prefix; default 4.

FourCC	Codec	CodecPrivateData
`H264`	H.264 / AVC	`00 00 00 01`, the sequence parameter set, `00 00 00 01`, the picture parameter set
`WVC1`	VC-1	A `VIDEOINFOHEADER` structure
`AACL`	AAC Low Complexity	Should be empty
`WMAP`	WMA Professional	A `WAVEFORMATEX` structure when `AudioTag` is 65534, otherwise empty
other	a code registered with the MP4 registration authority	Defined by its owner

AudioTag	Hex	Format
1	`0x0001`	PCM, 8 or 16 bits
85	`0x0055`	MP3
255	`0x00FF`	AAC
353	`0x0161`	WMA Standard (v7, v8, v9.x)
354	`0x0162`	WMA Professional (v9.x, v10): MS-SSTR lists it as 353 too, which is a slip; Windows’ wave format list gives 354
65534	`0xFFFE`	Extensible: `CodecPrivateData` should hold a `WAVEFORMATEXTENSIBLE` structure

CustomAttributes

Holds one or more Attribute elements, each with Name and Value. They tell apart tracks of the same stream that share a bit rate, and are added to fragment addresses as {CustomAttributes}.

Older IIS elements

Microsoft’s archived IIS documentation also lists, for text streams, a Content element with ScriptCommand (for SCMD), Marker (for CHAP) and Sample children. MS-SSTR does not define them; in it, text carried in the manifest goes in f elements of a stream with ManifestOutput. Their attributes are not documented in what remains of those pages.

Worked example

<SmoothStreamingMedia MajorVersion="2" MinorVersion="2" Duration="600000000">
  <StreamIndex Type="audio" Name="audio" Chunks="30" QualityLevels="1"
               Url="QualityLevels({bitrate})/Fragments(audio={start time})">
    <QualityLevel Index="0" Bitrate="128000" FourCC="AACL" AudioTag="255"
                  SamplingRate="48000" Channels="2" BitsPerSample="16" PacketSize="4"
                  CodecPrivateData="1190"/>
    <c t="0" d="20000000" r="30"/>
  </StreamIndex>
</SmoothStreamingMedia>

A 60-second audio-only presentation: thirty 2-second AAC fragments at 128 kbit/s. 1190 is the AAC AudioSpecificConfig for Low Complexity, 48 kHz, stereo, which encoders often supply although the specification says it may be empty.

.ISMC File Format – Manifest elements