Text-to-Speech
Piper
ONNX
Arabic
Algerian Arabic
tts
vits
algerian
darja
algerian-arabic

Hadra-TTS Piper-VITS: Algerian Darja Multi-Speaker ONNX Engine

Hadra-TTS Piper is an ultra-fast, offline neural text-to-speech engine optimized for embedded devices, Raspberry Pi, and mobile applications.

Voices & Speaker IDs

  • 0 (kahwa): Conversational Algerian podcast host with regional voiced velar realization (g).
  • 1 (rawi): Expressive male narrator voice (q).
  • 2 (loubna): Warm, melodic female storytelling voice (q).

Specifications

  • Architecture: VITS (Variational Inference with adversarial learning for end-to-end Text-to-Speech)
  • Format: ONNX Runtime (compatible with Piper C++ and Python CLI)
  • Sampling Rate: 22,050 Hz, 16-bit Mono PCM
  • Audio Latency: <50 ms on Raspberry Pi 4 / Modern CPU

Quickstart CLI Inference

echo "صحا خويا، راك en retard c'est bon، أرواح تشرب قهوة معانا." | piper   --model model.onnx   --config model.onnx.json   --speaker 0   --output_file output.wav
Downloads last month
9
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Datasets used to train touati-kamel/Hadra-TTS-piper