Dubbing

Knox exposes ElevenLabs-compatible dubbing endpoints at https://api.knoxstudio.ai/v1 for KnoxStudio. Authenticate with your Knox API key (Authorization: Bearer sk-... or xi-api-key). Knox bills billed operations and relays the request upstream. The same routes are also served by https://api.knox.chat/v1.

Dubbing takes source audio or video and produces a dubbed version in another language. It is not Voice Changer (same-language restyle) and not Text to Speech.

Typical flow:

1. List models — GET /v1/dubbing/models
2. Create a project — POST /v1/dubbing/project
3. Poll — GET /v1/dubbing/project/{project_id}
4. Fetch a language target — GET /v1/dubbing/project/{project_id}/language/{language_id}

https://api.knoxstudio.ai/v1
Authorization: Bearer sk-...

Use the official ElevenLabs SDK against Knox by pointing it at this base URL and using your Knox key as xi-api-key.

Models and billing

curl https://api.knoxstudio.ai/v1/dubbing/models

model_id	Use	Price
dubbing_v2	Default. Direct audio-to-audio dubbing, 90+ languages	$2.25 / source minute / language
dubbing_v1	Cascaded TTS dubbing, 70+ languages	$0.38 / source minute / language

Operation	Billing
Create project	First target language, billed on source duration
Add a language	Additional language, billed on source duration
Regenerate a target transcript	Billed on charged_seconds from the response
List / get / delete / transcript edits	Not billed

Source media may be up to 3 GiB. Provide exactly one of file or source_url. Default model is dubbing_v2.

Unauthenticated catalog requests are limited to 60/minute and may be cached for 5 minutes. A Bearer token returns only the models that token is allowed to use.

Create a project

source_language and target_language are BCP-47 tags (for example en, zh-CN). source_language is required when you attach a transcript. Transcript files are at most 4MB. At most 1000 keyterms (50 characters / 5 words each; <>{}[]\ are not allowed). At most 3 webhook_ids. Optional reference is at most 500 characters.

cURL

curl -X POST https://api.knoxstudio.ai/v1/dubbing/project \
     -H "Authorization: Bearer $KNOXCHAT_API_KEY" \
     -F "file=@episode.mp4" \
     -F "model_id=dubbing_v2" \
     -F "source_language=en" \
     -F "target_language=zh-CN"

Python

import os
import requests

with open("episode.mp4", "rb") as media:
    response = requests.post(
        "https://api.knoxstudio.ai/v1/dubbing/project",
        headers={"Authorization": f"Bearer {os.environ['KNOXCHAT_API_KEY']}"},
        files={"file": media},
        data={
            "model_id": "dubbing_v2",
            "source_language": "en",
            "target_language": "zh-CN",
        },
    )
response.raise_for_status()
print(response.json()["project_id"])

Project statuses: queued, preparing, processing, ready, failed. Language statuses: queued, processing, completed, stale, failed.

Transcripts and extra languages

Edit the source transcript, then the target translation, then regenerate:

Goal	Endpoint
Add a language	POST /v1/dubbing/project/{project_id}/language
Source transcript	GET /v1/dubbing/project/{project_id}/transcript
Add / update / delete source segments	/transcript/segment
Batch source updates	PATCH .../transcript/segments
Target transcript	GET .../language/{language_id}/transcript
Update target segments	PATCH .../transcript/segment/{segment_id}
Regenerate dubbed audio	POST .../transcript/regenerate

Source segments must last 0.1–25 seconds. Batch updates accept 1–500 segment entries. voice_settings.cloning_strength is 0–10.

Errors

422 uses { "detail": [{ "loc": [...], "msg": "...", "type": "value_error" }] }. Other errors use { "detail": "..." }.

Common statuses: 401 missing/invalid key, 402 insufficient credits, 403 model not allowed, 422 missing file/source_url, invalid BCP-47 language, or segment span out of range.