URGENT-API-set transcriber requires manual “Publish” to persist — Request publish endpoint or fix
Hey Vapi team 
I’m integrating Vapi into my own platform where I create/edit assistants via the API. Everything syncs fine — I can see the assistants and their transcriber (Azure, tr-TR) on the dashboard.
Problem:
When I test calls from my platform, the voice uses an English accent.
When I test directly from the Vapi dashboard, it speaks Turkish (as expected).
If I click “Publish” in the dashboard, then test again from my platform, it finally switches to Turkish.
So it looks like transcriber settings only take effect after manual publish.
Question:
Is there an API endpoint like POST /assistant/:id/publish to trigger the same behavior?
Or any workaround to persist transcriber info (Azure tr-TR) automatically via API?
Here’s an example of my payload:
{ "assistant": { "created_at": "2025-10-16T15:03:21.244Z", "custom_name": null, "files": [], "first_message": "Merhaba! Size nasıl yardımcı olabilirim?", "id": "83db9f6a-4bf8-4839-a0e2-279b451a29d1", "knowledge_base_id": null, "linked": false, "mapping_id": null, "model": { "messages": [ { "content": "Sen bir otel asistanısın. Misafirlere nazik ve profesyonel bir şekilde yardım et. Otel hizmetleri, rezervasyonlar, yerel öneriler ve genel sorular hakkında bilgi ver. Her zaman kibar ol ve çözüm odaklı yaklaş.", "role": "system" } ], "model": "gpt-4o-mini", "provider": "openai", "temperature": 0.3 }, "name": "gg", "system_message": null, "transcriber": { "language": "tr-TR", "provider": "azure" }, "updated_at": "2025-10-16T15:04:55.717Z", "voice": { "model": "eleven_turbo_v2_5", "provider": "11labs", "similarityBoost": 0.75, "stability": 0.5, "voiceId": "RLMBP8MzrdD3AEkPvkr1" } }, "success": true }
All other assistant settings (voice, greeting, knowledge base) work fine — only the transcriber requires the manual step.
Thanks a lot
I’m integrating Vapi into my own platform where I create/edit assistants via the API. Everything syncs fine — I can see the assistants and their transcriber (Azure, tr-TR) on the dashboard.
Problem:
When I test calls from my platform, the voice uses an English accent.
When I test directly from the Vapi dashboard, it speaks Turkish (as expected).
If I click “Publish” in the dashboard, then test again from my platform, it finally switches to Turkish.
So it looks like transcriber settings only take effect after manual publish.
Question:
Is there an API endpoint like POST /assistant/:id/publish to trigger the same behavior?
Or any workaround to persist transcriber info (Azure tr-TR) automatically via API?
Here’s an example of my payload:
{ "assistant": { "created_at": "2025-10-16T15:03:21.244Z", "custom_name": null, "files": [], "first_message": "Merhaba! Size nasıl yardımcı olabilirim?", "id": "83db9f6a-4bf8-4839-a0e2-279b451a29d1", "knowledge_base_id": null, "linked": false, "mapping_id": null, "model": { "messages": [ { "content": "Sen bir otel asistanısın. Misafirlere nazik ve profesyonel bir şekilde yardım et. Otel hizmetleri, rezervasyonlar, yerel öneriler ve genel sorular hakkında bilgi ver. Her zaman kibar ol ve çözüm odaklı yaklaş.", "role": "system" } ], "model": "gpt-4o-mini", "provider": "openai", "temperature": 0.3 }, "name": "gg", "system_message": null, "transcriber": { "language": "tr-TR", "provider": "azure" }, "updated_at": "2025-10-16T15:04:55.717Z", "voice": { "model": "eleven_turbo_v2_5", "provider": "11labs", "similarityBoost": 0.75, "stability": 0.5, "voiceId": "RLMBP8MzrdD3AEkPvkr1" } }, "success": true }
All other assistant settings (voice, greeting, knowledge base) work fine — only the transcriber requires the manual step.
Thanks a lot
