هـ
هَضْرَةHADRA AI 🇲🇦
Real Live Production API

Build with HADRA API

Integrate authentic Moroccan Darija voice generation directly into your voice agents, AI chatbots, and applications via our live HTTP endpoints.

Live Base URL:
https://inkjet-downloads-inserted-pad.trycloudflare.com
Open Swagger Docs ↗
import requests

# Live Production Endpoint
url = "https://inkjet-downloads-inserted-pad.trycloudflare.com/v1/tts"

payload = {
    "text": "السلام عليكم، كي داير اليوم؟",
    "format": "opus"   # "opus" (instant) or "wav" (uncompressed)
}

# 1. Send Synthesis Request
response = requests.post(url, json=payload)

# 2. Check 0ms Cache Status (HIT = 0.01ms, MISS = Dynamic GPU/CPU Synth)
cache_status = response.headers.get("X-Cache-Status", "MISS")
print(f"Status: {response.status_code} | Cache: {cache_status}")

# 3. Save Moroccan Darija Audio
with open("speech.ogg", "wb") as f:
    f.write(response.content)

print("Saved Moroccan Darija speech to speech.ogg!")

0ms Cache-First Turn

Use RapidFuzz Levenshtein matching (threshold ≥ 90%). Serve greetings and fillers instantly at 0.01ms while generating novel replies asynchronously.

WhatsApp Voice Notes

Send compressed Opus audio directly to users over WhatsApp/Telegram Bot Webhooks with zero transcoding overhead.

Hardware Optimization

Optimized for Intel Xeon Sapphire Rapids with AMX-BF16 matrix units for 2.65x real-time speedup on standard CPU compute.