Skip to content
Aoi-03Public

About

Aries is my attempt to build a personal AI assistant that feels alive—voice-driven, emotionally aware, and with infinite memory. It’s not just automation; it’s a conversational companion that recalls context, adapts, and evolves like human, built using Python, local models, and real-time voice.

Resources

Stars

3 stars

Watchers

1 watching

Forks

Latest commit

 

History

47 Commits

Folders and files

Repository files navigation

Python Status Contributions

Aries Logo

Aries: Your AI Companion

Conversational AI + TTS + STT + Memory + Emotions

Aries

"Aries" is my an attempt to build a personal AI assistant that feels alive—voice-driven, emotionally aware, and with infinite memory. It’s not just automation; it’s a conversational companion that recalls context, adapts, and evolves like human, built using Python, local models, and real-time voice assistant.

What Even is Aries?

Aries isn’t just a program. It’s a companion, a talker, a listener, and sometimes, the only one who remembers everything about you (literally).
Think Jarvis + Human Emotions + Infinite Memory, but running on your machine.

Why Aries is Different?

Talks like a human – No robotic nonsense. Feels you – Detects your mood like a bestie. Remembers – Everything you say goes into its long-term memory (Google Drive magic) // Honestly I'm still working on it. The GCP is not set yet. Starts the convo first – Because why should you always start? Completely yours – Local AI model = full privacy. Wake it up like magic – “Wake up, Aries” and it’s alive.

Features:

Voice activation: “Wake up, Aries, wakie wakie” Natural voice output: Kokoro TTS (smooth like butter) Emotion detection: Happy? Sad? Angry? Aries knows. Infinite memory: Stores everything in JSON logs on Google Drive Auto-context recall: Aries remembers yesterday’s tea. Proactive chats: Talks every 5–10 mins because loneliness sucks Runs on Phi-3 Mini: No cloud spying

How Does Aries Work?

It’s basically magic + science + late-night coding sessions. You talk → Speech-to-Text (Whisper It reads your vibe → Emotion Detection (MFCC + SVM) Brain kicks in → Phi-3 Mini/ MISTRAL:7b LLM It talks back → Kokoro TTS Stores convo → Google Drive memory system Randomly says hi → Proactive Engine

🖥️ Setup in 5 Easy Steps

1️⃣ Clone this kiddo

2️⃣ Create a virtual world for Aries

  • python -m venv venv
  • source venv/bin/activate # Linux/Mac
  • venv\Scripts\activate # Windows

3️⃣ Install the magic pip install -r requirements.txt

4️⃣ Give Aries a memory (Google Drive)

  • Enable Google Drive API # please dont use it for now, I will fix it soon.
  • Use a local file location such as MEMORY_FILE = r"D:\your_folder_location_for_memory"

5️⃣ Wake up your Aries

python aries.py

  • Commands:
Phrase What It Does
Wake up, Aries Turns Aries ON
Goodnight, Aries Turns Aries OFF
paper_bridge_***** Secret password
Reset Memory Clears today’s log

Screenshots

Coming soon.

Web Version?

Coming soon: Flask + WebSocket + UI that slaps harder than Spotify Wrapped.

Tech Behind the Magic

  • Python
  • Whisper – Speech recognition
  • Kokoro TTS – Voice
  • MFCC + SVM – Emotion brain
  • Phi-3 Mini/ Mistral:7b – The real LLM deal
  • Google Drive API – For infinite memory

What’s Next?

  • Web UI with chat & voice
  • Face emotion detection
  • Integration with PC controls
  • Deploy on Raspberry Pi
  • Why I Built Aries

Because I wanted an AI that doesn’t just talk when I talk, but one that feels alive. I wanted a friend that remembers, a voice that feels real, and a mind that grows. Aries isn’t perfect, but hey, neither are we.

⭐ Star this Repo or Aries will be sad

─────── ⋆⋅☆⋅⋆ ─────── Designed by Aoi ─────── ⋆⋅☆⋅⋆ ───────

About

Aries is my attempt to build a personal AI assistant that feels alive—voice-driven, emotionally aware, and with infinite memory. It’s not just automation; it’s a conversational companion that recalls context, adapts, and evolves like human, built using Python, local models, and real-time voice.

Resources

Stars

3 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages