Using HLS inputs with MediaConvert

When your input to MediaConvert is an HLS package, specify either a parent or child manifest for Input file URL (FileInput). When the manifest is a parent that lists multiple child manifests, MediaConvert uses the child manifest with the highest bandwidth as the input source.

Features compatible with HLS inputs

With HLS inputs, you can use the following input features:

• Input clipping

• Input stitching

• Image insertion

• Embedded input captions selectors

Feature restrictions with HLS inputs

When your input is an HLS package, your job is restricted in these ways:

• Your input package must conform to the requirements listed in HLS input package requirements.

• Your input segments can't be encrypted with DRM. For example, your inputs can't be encrypted with Apple FairPlay DRM.

• You can use only embedded input captions.

HLS input package requirements

Your input HLS package must conform to the following requirements:

• The video container for your media segments must be MPEG-2 TS.

• The compatibility version of the manifest file (specified by EXT-X-VERSION) must be 4 or smaller.

• The manifest file must stay the same after you submit your job. That is, the manifest must have the tag EXT-X-ENDLIST or it must have the EXT-X-PLAYLIST-TYPE value set to VOD.

• If the manifest uses EXT-X-BYTERANGE, the start of the first subrange must be 0 and the following subrange segments must continue the former one.

• If the input has discontinuities, they must start at the beginning of a segment. That is, the input can't have discontinuities in the subrange of a segment.

• If the manifest includes EXT-X-KEY, METHOD must be set to NONE. For example: EXT-X-KEY:METHOD=NONE. MediaConvert does not support HLS encrypted inputs.

• MediaConvert ignores the following tags:

  • EXT-X-PROGRAM-DATE-TIME

  • EXT-X-DATERANGE

  • EXT-X-I-FRAMES-ONLY

  • EXT-X-I-FRAME-STREAM-INF

  • EXT-X-SESSION-DATA

  • EXT-X-SESSION-KEY

  • EXT-X-INDEPENDENT-SEGMENTS

  • EXT-X-START

• When your job uses accelerated transcoding, your input HLS package must conform to this additional requirement: The duration in EXTINF must be specified using a decimal floating-point, with enough accuracy to avoid perceptible errors when segment durations are accumulated.

Using alternate audio renditions

With HLS rendition groups, you can use the audio selector settings to indicate which alternate audio rendition you want MediaConvert to use. To be eligible for selection, your alternate audio renditions must conform to the following requirements:

• The renditions must be included in EXT-X-MEDIA tags in the input parent manifest.

• The EXT-X-MEDIA tags must contain a unique combination of GROUP-ID, NAME, and LANGUAGE values..

• Audio must be in one of the following supported audio codecs: AAC, Dolby Digital (AC3), Dolby Digital Plus (EAC3), or MP3.

• The child manifest for your alternate audio rendition must be included in the parent manifest that you used for your Input file URL (FileInput)

When you specify audio selector settings to identify an alternate audio rendition, the audio selector looks for a matching EXT-X-MEDIA tag in the parent manifest.

You can use one or more selector settings at a time. For example, given the following EXT-X-MEDIA tags, you can identify the audio rendition by the name (RenditionName) or language (RenditionLangageCode) because these are both unique values across the tags.

#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio",CHANNELS="2",NAME="English",LANGUAGE="eng",DEFAULT=YES,AUTOSELECT=YES,URI="english_audio.m3u8"

#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio",CHANNELS="2",NAME="Japanese",LANGUAGE="jpn",DEFAULT=NO,AUTOSELECT=NO,URI="japanese_audio.m3u8"

However, because the group ID (RenditionGroupID) is the same for both tags, you can't use that alone to identify an audio rendition. You must use the group ID in combination with another value from the EXT-X-MEDIA tag to identify the audio rendition that you want MediaConvert to use.

If you don't specify audio selector settings, the audio selector looks for audio that's muxed into the video segments. If the video segments don't contain audio, the audio selector uses the first alternate audio rendition from the input parent manifest.