Concepts

Which Model Should I Use?

You can choose your path depending on how much control you want:

✅ I know the model

Call a specific endpoint:

/tts/vui
/tts/orpheus
/tts/koroko
/tts/xtts/v2
/stt/whisper-v3
/stt/whisper-v3-turbo
/dia/whisperx

Use this when:

You want precise control over latency or voice
You're doing A/B testing
You’re optimizing by region or feature set

🤖 I have requirments

Call:

/tts
/stt

We’ll route your request to the best model based on:

Language
Region
Pricing limit
Cloning requirement
Quality score

Perfect if you want to start fast or stay flexible.

🔊 TTS Models

Endpoint	Model	Description
`/tts/vui`	VUI	Fast, low-latency voice model with natural prosody. Great default.
`/tts/orpheus`	Orpheus	Richer expressive tone; good for more emotional or human-like delivery.
`/tts/koroko`	Koroko	Multi-language support with speaker control and voice cloning options.
`/tts/xtts/v2`	XTTS v2	High-fidelity multilingual model with speaker ID, cloning, and accent precision.
`/tts/mars6`	Mars 6 (optional)	Experimental or internal-use model; not always available.

🎤 STT Models

Endpoint	Model	Description
`/stt/whisper-v3`	Whisper v3	Accurate, multilingual transcription with broad language support.
`/stt/whisper-v3-turbo`	Whisper Turbo	Cost-effective and faster version for bulk or low-latency tasks.
`/dia/whisperx`	WhisperX	Diarization-capable transcription with word-level timestamps.

📦 Example: Smart-Routed

Code(bash)
 
curl -X POST https://api.slng.ai/tts \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "text": "Smart routing in action", "language": "en-US" }'

📦 Example: Explicit Model

Code(bash)
 
curl -X POST https://api.slng.ai/tts/vui \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "text": "Direct call to VUI" }'

Last updated: June 2025

Last modified on July 21, 2025

Getting Started Setting Preferences