Generative Audio API: TTS, Voice Cloning & Speech Recognition

Generative Audio API Overview

Hypereal provides a comprehensive suite of audio generation APIs for text-to-speech, voice cloning, and speech recognition. All audio models are accessible through a unified API with competitive pricing.

Available Audio Models

Model	Slug	Description	Pricing
Text to Speech	audio-tts	High-quality TTS with 64+ emotions	$0.015/1000 chars
Voice Clone	audio-clone	Zero-shot voice cloning	$0.015/1000 chars
Speech Recognition	audio-asr	Transcribe audio to text	$0.006/minute
Minimax Voice Clone	minimax-voice-clone	Premium voice cloning	$0.50/generation
Speech Turbo	minimax-speech-02	Fast TTS with emotion control	$0.003/generation
Music Generation	minimax-music-02	AI music with vocals	$0.045/song

Emotional Text-to-Speech

One of the most powerful features is emotion control with 64+ emotional expressions organized into four categories.

Basic Emotions (24)

Core emotional states for natural speech:

happy, sad, angry, excited, calm, nervous, confident, surprised, satisfied, delighted, scared, worried, upset, frustrated, depressed, empathetic, embarrassed, disgusted, moved, proud, relaxed, grateful, curious, sarcastic

Advanced Emotions (25)

More nuanced expressions:

disdainful, unhappy, anxious, hysterical, indifferent, uncertain, doubtful, confused, disappointed, regretful, guilty, ashamed, jealous, envious, hopeful, optimistic, pessimistic, nostalgic, lonely, bored, contemptuous, sympathetic, compassionate, determined, resigned

Tone Markers (5)

Delivery style modifiers:

• in a hurry tone - Urgent delivery
• shouting - Loud, emphatic
• screaming - Intense, high volume
• whispering - Soft, intimate
• soft tone - Gentle delivery

Audio Effects (10)

Sound effects and vocalizations:

laughing, chuckling, sobbing, crying loudly, sighing, groaning, panting, gasping, yawning, snoring

Plus special effects: audience laughter, crowd laughter, pause breaks.

Emotion Syntax

Wrap emotions in parentheses at the start of your text:

(happy) What a beautiful day!
(sad) I'm sorry for your loss.
(excited) I can't believe we won!

Combining Emotions

Stack multiple tags for complex expressions:

(sad)(whispering) I'll miss you.
(excited)(laughing) This is amazing!
(nervous)(in a hurry tone) We need to go now!

API Examples

Text-to-Speech with Emotion

const response = await fetch('https://api.hypereal.com/v1/audio/generate', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer YOUR_API_KEY',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: 'audio-tts',
    text: '(excited) Welcome to our platform! We are so happy to have you here.',
    format: 'mp3',
    temperature: 0.7
  })
});

Voice Cloning

const response = await fetch('https://api.hypereal.com/v1/audio/generate', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer YOUR_API_KEY',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: 'audio-clone',
    text: '(confident) This is my cloned voice speaking.',
    audio: 'https://example.com/my-voice-sample.mp3',
    format: 'mp3'
  })
});

Speech Recognition

const response = await fetch('https://api.hypereal.com/v1/audio/generate', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer YOUR_API_KEY',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: 'audio-asr',
    audio: 'https://example.com/speech-recording.mp3',
    language: 'en',
    ignore_timestamps: false
  })
});

// Response includes text and timestamps
// { text: "...", duration: 5.2, segments: [...] }

Best Practices

Emotion Usage

1. Start simple - Test with basic emotions before combining
2. Match content - Align emotions with your text meaning
3. Don't overuse - Avoid too many emotion tags in short text
4. Test variations - Different voices may express emotions differently

Voice Cloning

1. Quality reference - Use clean, noise-free audio (10-30 seconds)
2. Clear speech - Reference should have clear pronunciation
3. Enable enhancement - Use enhance_audio_quality: true for noisy samples

Speech Recognition

1. Specify language - Improves accuracy significantly
2. Quality audio - Clear recordings produce better results
3. Use timestamps - Enable for subtitle/caption generation

Output Formats

All TTS and voice cloning models support:

• MP3 - Best for general use (default)
• WAV - Uncompressed, best for editing
• PCM - Raw audio data
• Opus - Efficient for streaming

Supported Languages

Multi-language support including:

• English (en)
• Chinese (zh)
• Japanese (ja)
• Spanish (es)
• French (fr)
• German (de)
• And many more

Why Choose Hypereal for Audio?

• Unified API - One endpoint for TTS, cloning, and ASR
• Competitive pricing - Up to 80% cheaper than alternatives
• 64+ emotions - Industry-leading expression control
• Zero-shot cloning - Clone any voice from a short sample
• Fast processing - Optimized for low latency

Get Started Free - No credit card required.