← All models
PersonaPlex is a real-time, full-duplex speech-to-speech conversational model that enables persona control through text-based role prompts and audio-based voice conditioning.
Details
- Accepts
- audio
Pricing
- Price
- 0.125 cr / second
Prices in credits (1 credit = $0.01).
Data schema
Input
| Field | Type | Description |
|---|---|---|
| seed | — | Random seed for reproducibility. |
| audiorequired | string | Input audio chunk (PCM s16le, 24kHz mono). Base64-encoded in JSON transport. |
| voice | stringenum: NATF0, NATF1, NATF2, NATF3, NATM0, NATM1… | Voice ID for the AI response. NAT = natural, VAR = variety. F = female, M = male. Ignored when voice_audio_url is provided. |
| prompt | string | Text prompt describing the AI persona and conversation context. |
| top_k_text | integer | Top-K sampling for text tokens. |
| top_k_audio | integer | Top-K sampling for audio tokens. |
| voice_audio_url | — | URL to a voice sample audio for on-the-fly voice cloning. When provided, the AI responds in the cloned voice instead of the preset 'voice'. 10+ seconds of clear speech recommended. Billed at 2x rate. |
| temperature_text | number | Text sampling temperature. Higher values produce more diverse outputs. |
| temperature_audio | number | Audio sampling temperature. Higher values produce more diverse outputs. |
Output
| Field | Type | Description |
|---|---|---|
| text | string | Generated text tokens for this chunk. |
| audio | string | Generated audio chunk (PCM s16le, 24kHz mono). Base64-encoded in JSON transport. |