Model Information

Display Name: HiVox TTS

API Model ID: hinow/hivox-tts

Category: Text To Audio

Description: HiVox TTS turns text into speech and starts talking before the sentence finishes generating. Built for voice agents, where waiting for a full audio file is the difference between a conversation and an awkward pause. **Key Features:** - Real-time streaming: first audio in under a second - Brazilian Portuguese voices, plus English, Spanish, French and German - Voice cloning: bring your own voice - Automatic failover between upstream engines - Billed by minute of generated audio, not by character **Capabilities:** - Live voice agents and phone-style conversations - Narration and audio replies inside chat flows **Best For:** - Voice agents in Portuguese - Any flow where time-to-first-audio is what the user feels **Technical Specs:** - Formats: pcm (default, lowest latency), wav, opus, mp3 - Sample rate: 48 kHz mono - Billed by measured audio duration

How to Use This Model

To use HiVox TTS via the HInow.ai API, use the model ID: hinow/hivox-tts

API Request Example (Chat/Text)

POST https://api.hinow.ai/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "model": "hinow/hivox-tts",
  "messages": [
    {"role": "user", "content": "Your message here"}
  ]
}
              

Pricing

• per_minute: $0.05

Available Parameters

• voice: Voice name. Bianca and Mateus speak Brazilian Portuguese. (Options: Bianca, Mateus, Harper, Sterling, Vera, Mateo, Solene, Marius, Lorena, Mats)
• voice_id: Your own cloned voice id. Takes precedence over voice.
• language: Language. Picks the default voice for it when none is given. (Options: pt, en, es, fr, de)
• response_format: pcm has the lowest latency: raw samples, nothing to decode. (Options: pcm, wav, opus, mp3)
• speed: Speech speed multiplier.

Quick Reference

To use this model, set: "model": "hinow/hivox-tts"

Featured: Yes

Documentation: https://hinow.ai/models/hinow/hivox-tts

API Endpoint: https://api.hinow.ai/v1