Kitta AI Docs
Integration GuidesThird-Party Integrations
Back to site

OpenTalking Setup Guide

Configure the OpenAI-compatible TTS provider in OpenTalking to use a Kitta AI model and cloned voice.

Connect Kitta AI TTS to OpenTalking

OpenTalking includes an openai_compatible TTS provider for voice conversations and generated media. Its TTS provider is separate from its LLM and speech-recognition providers.

Prepare the API key and Voice ID

  1. Create a dedicated key in API management.
  2. Copy the Voice ID from your cloned voice.
  3. Confirm that your API credit balance is sufficient.

Use this system voice for the first test:

00a1b221-6137-4b73-ad62-b0cbce134167

Configure OpenTalking

Add or update these values in OpenTalking's backend .env file:

OPENTALKING_TTS_DEFAULT_PROVIDER=openai_compatible
OPENTALKING_TTS_OPENAI_BASE_URL=https://kittaai.com/v1
OPENTALKING_TTS_OPENAI_API_KEY=YOUR_FISH_AUDIO_API_KEY
OPENTALKING_TTS_OPENAI_MODEL=fishaudio-s21pro-flash
OPENTALKING_TTS_OPENAI_VOICE=00a1b221-6137-4b73-ad62-b0cbce134167
OPENTALKING_TTS_OPENAI_RESPONSE_FORMAT=wav
OPENTALKING_TTS_OPENAI_PROTOCOL=audio_speech

If OPENTALKING_TTS_ENABLED_PROVIDERS is already set, make sure its comma-separated list includes openai_compatible.

Use the Base URL ending at /v1

OpenTalking appends /audio/speech to OPENTALKING_TTS_OPENAI_BASE_URL. Do not put the complete /v1/audio/speech endpoint in this variable.

Restart the OpenTalking backend after changing .env. In Settings, choose OpenAI-compatible TTS, enter a short preview sentence, and play the voice preview.

Provider boundaries

  • These settings change TTS only. Configure the LLM and ASR providers separately.
  • Keep audio_speech as the protocol. Kitta AI is not a chat-completions audio provider.
  • Use wav for the first test because OpenTalking's adapter decodes WAV directly before streaming audio to the conversation.
  • Replace the system Voice ID with your cloned voice only after the preview works.

Troubleshooting

  • 404: use https://kittaai.com/v1 as the Base URL; OpenTalking adds /audio/speech.
  • 401: verify OPENTALKING_TTS_OPENAI_API_KEY, then restart the backend.
  • 402: add API credits before trying again.
  • 400 or invalid voice: use fishaudio-s21pro-flash and a valid Kitta AI Voice ID.
  • OpenAI-compatible TTS is unavailable: add openai_compatible to OPENTALKING_TTS_ENABLED_PROVIDERS and restart.
  • Settings changed but audio did not: OpenTalking persists runtime configuration to the backend environment; confirm the displayed model and voice, then restart the affected service.

AIRI Setup Guide

Configure Kitta AI as an OpenAI-compatible speech provider in AIRI and assign a cloned voice to an AI character.

Android Custom TTS Setup Guide

Connect Custom TTS Service for Android to Kitta AI and use a cloned voice in apps that support the Android system TTS engine.

On this page

Connect Kitta AI TTS to OpenTalkingPrepare the API key and Voice IDConfigure OpenTalkingProvider boundariesTroubleshootingRelated documentation