Go to https://github.com/fal-ai/fal/issues/new and use something like this: Title: text-to-dialogue/eleven-v3 returns 422 "Voice not found" even with default playground example

[AbhishekDhoundiyal]

Endpoint: fal-ai/elevenlabs/text-to-dialogue/eleven-v3

Issue: The endpoint returns a 422 with the following error, even when using the default example on the fal playground:

{
    "detail": [
        {
            "loc": ["body", "voice"],
            "msg": "Voice not found: ",
            "type": "feature_not_supported",
            "url": "https://docs.fal.ai/errors#feature_not_supported",
            "input": ""
        }
    ]
}

Steps to reproduce:

1. Go to https://fal.ai/models/fal-ai/elevenlabs/text-to-dialogue/eleven-v3
2. Reset to default values
3. Hit Run
4. Get 422 error

Notes:

• The pre-cached default output displays fine, but any actual Run request fails
• Single TTS endpoint (fal-ai/elevenlabs/tts/eleven-v3) works perfectly
• Kling video endpoints work perfectly
• The error suggests a top-level voice field is being sent as empty string, which doesn't exist in the dialogue schema (voices are per-input)
• Tested via both playground and API with fal_client

Environment: fal_client (Python), also reproducible directly on playground