mayaresearch/veena-max

VeenaMAX is a high-performance, advanced Text-to-Speech API that converts text into natural-sounding speech with emotional intelligence and blazing-fast response times. It can speak English and Hindi - the most widely used languages in India. As a signifi

Public
25 runs
Run with an API
Pricing
License
Weights
Playground API Examples README Versions

State-of-the-art TTS system for high-quality Hindi and bilingual (Hindi-English) speech synthesis. Built to deliver natural-sounding voices with emotional depth and clarity.

Available Voices

Female Voices

  • vinaya_assist - Professional assistant voice (default)
  • charu_soft - Gentle storyteller’s voice
  • keerti_joy - High-energy, vibrant tone
  • mohini_whispers - Whisper-soft, ASMR optimized
  • maitri_connect - Expressive narration voice
  • soumya_calm - Smooth, soothing voice

Male Voices

  • varun_chat - Deep, authoritative voice
  • agastya_impact - Bold, confident presence

Parameters

Parameter Default Range Description
text (required) Min 30 chars Text to convert to speech
speaker vinaya_assist See voices above Voice selection
temperature 0.5 0.0-2.0 Controls randomness
repetition_penalty 1.1 1.0-2.0 Reduces repetition
top_p 0.9 0.0-1.0 Nucleus sampling
max_new_tokens 800 100-4096 Token limit
normalize_text true true/false Auto text normalization
seed null Any integer For reproducible output
output_format wav wav/opus/webm Audio format

Use Cases

  • Audiobooks & Narration - Use charu_soft for storytelling
  • Virtual Assistants - Use vinaya_assist for professional responses
  • News & Announcements - Use varun_chat for authoritative delivery
  • ASMR & Meditation - Use mohini_whispers for calming content
  • Educational Content - Use keerti_joy for engaging presentations
  • Corporate Presentations - Use agastya_impact for impactful delivery

Tips

  1. Text Length: Optimal 50-500 characters per request
  2. Consistency: Use temperature 0.2-0.4 for consistent output
  3. Reproducibility: Set a seed value for identical outputs
  4. Language: Supports pure Hindi, pure English, and code-mixed text
  5. Numbers: Automatically normalized (e.g., “2024” → “two thousand twenty four”)

Language Features

  • ✅ Full Devanagari script support
  • ✅ Natural English pronunciation
  • ✅ Seamless Hindi-English code-mixing
  • ✅ Automatic number/date normalization
  • ✅ Proper punctuation handling

Output Formats

  • WAV: Uncompressed, highest quality (default)
  • Opus: Compressed, excellent for streaming
  • WebM: Compressed, web-optimized