Skip to content
YA
All work
PrototypeAndroidAI

Sadaصدى

An Arabic voice assistant designed to run fully offline — speech in, speech out.

Status
v0.1: LLM, TTS, audio and data layers working; Whisper STT is a documented stub.
Period
2026
Platforms
Android
Repository
Private — available on request
Live
—
Sصدى

Sada

  • AudioRecorder 16 kHz PCM
  • WhisperEngine (stub)
  • LlmEngine (Gemma)
  • SadaTts
  • SonicOrb (4 states)
  • conversation
  • history
  • settings

01Overview

The second app in an on-device AI series after Qandeel: live recording with a reactive waveform, Gemma 3 1B replies on the device and Arabic speech via Android TTS, with a four-state 'SonicOrb'. Speech recognition through whisper.cpp is designed but still a stub in v0.1 — documented as the first task of v0.2.

02Screens

No screenshots exist for this project yet, so it is presented through its architecture rather than invented screens.

03System architecture

  1. L1Voice loop
    • AudioRecorder 16 kHz PCM
    • WhisperEngine (stub)
    • LlmEngine (Gemma)
    • SadaTts
  2. L2App
    • SonicOrb (4 states)
    • conversation
    • history
    • settings
Audio → STT → LLM → TTS, each behind its own engine interface.Diagram — derived from the source code

Technology

  • Kotlin
  • Jetpack Compose
  • MediaPipe
  • Gemma
  • Whisper
  • Android TTS
  • Hilt
  • Room
  • On-device LLM

How this was verified

Source and README reviewed, including its own readiness table.

Related projects

Next project

Nutq