NouAI Voice Synthesis API
Integrate ultra-realistic Lao and Thai speech directly into your apps, bots, backend services, and workflows with simple HTTP REST requests.
AI-Ready DocumentationLLM Friendly
Need an AI agent to write the integration code for you? Click the button to copy the full Markdown specification.
TTS Models & Character Limits
Choose between NouAI Speech v2 (fast & expressive) or NouAI Speech v1 (extended 4,000 characters long-form). Both cost flat 30 credits.
NouAI Speech v2nouai-speech-v2
Breeze TTS 2 EngineHigh-fidelity conversational voice synthesis with 20 studio voices and expressive vocal event tags (like laughing, sighing).
NouAI Speech v1nouai-speech-v1
OmniVoice EngineEngineered for long-form scripts, news broadcasts, storytelling, and educational content. Synthesizes up to 4,000 characters in a single request!
curl -X POST https://nouai.app/api/v1/tts/speak \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"text": "Nyob zoo sawv daws, zoo siab txais tos nej tuaj rau NouAI Studio.",
"model": "nouai-speech-v2",
"voice": "alice",
"speed": 1.0
}' \
--output speech.wavRequest Parameterslanguage is automatic
Endpoint: POST /api/v1/tts/speak • Flat 30 credits per request
NouAI Speech v2nouai-speech-v2
Breeze TTS 2 EngineOptimized for natural conversational speech, voicebots, and dialogue. Supports 20 studio voices and inline vocal tags like (laugh), (sigh).
| Field | Type | Req | Description |
|---|---|---|---|
| text | string | Yes | Text to synthesize. Max 1,500 chars. |
| model | string | No | "nouai-speech-v2" (default if omitted). |
| voice | string | No | Preset voice: alice, aria, sarah, daniel, charlie, etc. (20 studio voices). |
| speed | number | No | Playback speed multiplier (0.5 to 2.0). Default 1.0. |
NouAI Speech v1nouai-speech-v1
OmniVoice EngineEngineered for long articles, news reading, audiobooks, and storytelling. Allows up to 4,000 characters per single synthesis request without cutting.
| Field | Type | Req | Description |
|---|---|---|---|
| text | string | Yes | Text to synthesize. Max 4,000 chars for long-form context. |
| model | string | Yes | Must specify "nouai-speech-v1" to enable 4,000 chars. |
| voice | string | No | Preset voice: default_voice, female, female_2, teenneger, teenneger_2. |
| speed | number | No | Playback speed multiplier (0.5 to 2.0). Default 1.0. |
Audio Reference Parameters (ref_audio)
Provide 3–15 seconds of clean speech audio to dynamically clone and mimic any speaker's voice. When reference audio is provided, it automatically overrides the preset voice parameter.
| Field | Type | Req | Description |
|---|---|---|---|
| ref_audio_base64 | string | No* | Base64-encoded audio bytes of the target speaker (WAV, MP3, FLAC, M4A, OGG). Max 10MB. 3–15 seconds of clear speech recommended. |
| ref_audio_suffix | string | No | Audio file extension: ".wav" (default), ".mp3", ".flac", ".m4a", ".ogg". |
| ref_audio | string | No* | Direct public URL (https://...) or Data URI (data:audio/wav;base64,...) pointing to reference audio. Alternative to ref_audio_base64. |
| ref_text | string | No | Exact transcript text of words spoken in the reference audio. (Highly recommended — improves voice similarity and pronunciation). |
ref_audio_base64 or ref_audio to activate voice cloning.Billing: 30 credits flatResponse Headers & Status Codes
| Header | Description |
|---|---|
| Content-Type | audio/wav binary audio stream. |
| X-Audio-Duration-Seconds | Total audio length in seconds. |
| X-Credits-Charged | Credits deducted (always 30). |
| X-Credits-Remaining | Remaining account balance. |
| X-RateLimit-Remaining | Calls remaining in current 1-min window. |
Live API Tester
Test synthesis directly from your browser. Flat 30 credits deducted per call.