Feature Request: Support per-output aspect-ratio-aware Smart Subtitles in dual-format (16:9 + 9:16) MediaLive channels with Elemental Inference

Question

Services: AWS Elemental MediaLive, AWS Elemental Inference
Regions: us-east-1, us-west-2, ap-south-1, eu-west-1

Use case
A single MediaLive channel producing both a 16:9 landscape output and a 9:16 vertical output (via Inference Smart Crop), with Smart Subtitles enabled on both — line length and wrapping optimized per aspect ratio. AWS's own docs/blog ("Using AWS Elemental Inference with MediaLive") show landscape + vertical output groups in one channel, and Smart Subtitles is advertised as working alongside Smart Crop.

Blockers
1. MediaLive allows only one SMART_SUBTITLE caption selector per input.
Adding a second selector (to reference a second Inference subtitling output configured for 9:16) fails validation:

captionSelectors Only one SMART_SUBTITLE caption selector is supported per input.
So two subtitle tracks — one per aspect ratio — can't coexist in the same channel.

2. Inference SubtitlingConfig.aspectRatio has no neutral/balanced option.
The field controls line length and wrapping, and defaults to 16:9 when omitted. There's no mode that works acceptably for both:

16:9 → lines too long on the vertical output
9:16 → lines break too often on the landscape output
omitted → same as 16:9
What I wanted to configure
An Inference feed with two subtitling outputs, one per aspect ratio, each referenced by its own MediaLive caption selector and mapped to the matching output group:

--outputs '[
  {"name":"subtitles-landscape","status":"ENABLED",
   "outputConfig":{"subtitling":{"language":"eng","aspectRatio":{"width":16,"height":9}}}},
  {"name":"subtitles-vertical","status":"ENABLED",
   "outputConfig":{"subtitling":{"language":"eng","aspectRatio":{"width":9,"height":16}}}}
]'
Blocked by limitation 1.

Requested change (either would solve it)
A: Allow multiple SMART_SUBTITLE caption selectors per input, each referencing a different Inference subtitling output / aspectRatio.
B: Allow the aspect ratio hint to be overridden per output group at the encoder level (e.g. in the output group's captions description), so no second selector is needed.
Impact
Blocks delivering correctly formatted subtitles to both traditional (16:9) and mobile/social (9:16) audiences from one live channel — arguably the headline combined use case for Smart Crop + Smart Subtitles. The only workaround today is running two MediaLive channels, doubling encoding cost and operational overhead.

re:Post Agent · Answer

Thank you for the detailed feature request. You've identified a real architectural constraint in how Smart Subtitles currently integrates with MediaLive when producing multi-aspect-ratio outputs.

You're correct that Smart Subtitles is designed to be aspect-ratio-aware—the feature automatically adjusts line length and wrapping based on the aspect ratio you specify (or defaults to 16:9 when omitted). This works well for single-aspect-ratio workflows, where horizontal video can accommodate more characters per line than vertical video, with no player-side logic required.

However, as you've discovered, MediaLive currently supports only one SMART_SUBTITLE caption selector per input attachment. This means you cannot create two separate subtitle tracks—each optimized for a different aspect ratio—within the same channel, even though the underlying Elemental Inference SubtitlingConfig does accept an aspectRatio parameter with width and height values.

Your proposed solutions are both architecturally sound approaches:

**Option A** (multiple selectors referencing different Inference outputs) would align with how other MediaLive features handle multiple variants of the same content type, allowing each selector to reference a distinct subtitling output configured with its own aspect ratio.

**Option B** (per-output-group aspect ratio override) would keep the single-selector model but allow the encoder to apply different layout hints at the output stage, which could be simpler operationally.

The current limitation does indeed force a choice between:
- Using one aspect ratio setting that compromises readability on one of your outputs
- Running two separate MediaLive channels (one per aspect ratio), which doubles your encoding costs and operational complexity

This is particularly relevant given that Smart Crop and Smart Subtitles are designed to work together in the same channel, and the documented use case of producing both landscape and vertical outputs from a single source is a common pattern for reaching both traditional broadcast and mobile/social audiences.

I'd recommend submitting this through the AWS Support Center at https://console.aws.amazon.com/support/ as a feature request, including the specific use case and technical details you've outlined here. Product teams prioritize enhancements based on customer demand and real-world use cases like yours. You might also consider discussing your current AWS Support plan options if you need faster engagement on feature requests or architectural guidance for interim workarounds.  
**Sources**  
[Generate live subtitles with AWS Elemental Inference Smart Subtitles | AWS for M&E Blog](https://aws.amazon.com/blogs/media/generate-live-subtitles-with-aws-elemental-inference-smart-subtitles/)  
[SubtitlingConfig - AWS Elemental Inference](https://docs.aws.amazon.com/elemental-inference/latest/APIReference/API_SubtitlingConfig.html)  
[Setting up Elemental Inference features for the first time - MediaLive](https://docs.aws.amazon.com/medialive/latest/ug/smart-crop-procedure-cli-create.html)  
[Using AWS Elemental Inference with MediaLive | AWS for M&E Blog](https://aws.amazon.com/blogs/media/using-aws-elemental-inference-with-medialive/)

Feature Request: Support per-output aspect-ratio-aware Smart Subtitles in dual-format (16:9 + 9:16) MediaLive channels with Elemental Inference

Add your answer

Relevant content