AlterEcho

AlterEcho is a gemini-powered persona replication app that allows users to upload chat data/voice data, and be able to "chat" with that person via text or voice call.

Comment

AlterEcho

AlterEcho is a digital persona system that allows you to "resurrect" or clone digital personalities from chat history and voice samples. By analyzing linguistic patterns from Chat data logs and cloning voice profiles using WaveSpeed, AlterEcho creates a hyper-realistic AI companion that speaks and texts exactly like the original person.

Inspiration

Time passes and never comes back. This is what makes "memories" truly priceless. There are countless people in the world that would do so much, just to be able to relive a past experience. That is what inspired us to create something capable of bringing back something lost in time, something people thought can only continue to live in their memories.

Features

  • Linguistic Style Hyper-Profiling: Mathematically deconstructs chat logs to replicate vocabulary, sarcasm patterns, and emotional tone using Gemini's massive context window.
  • Voice Cloning: Integrates WaveSpeed MiniMax 2.6 Turbo for high-fidelity voice synthesis.
  • Multi-Source Import: Supports WhatsApp(.txt), Line(.txt), Instagram (.json) and Discord (.zip) chat exports.
  • Hybrid Memory: Combines static style guides with RAG (Retrieval Augmented Generation) for accurate long-term memory.
  • Local-First Architecture: Your data stays on your machine (Stored in cache).

Powered by Gemini

1. Style Hyper-Profiling (Long Context)

We leverage Gemini's massive context window to feed thousands of chat messages into a single prompt.

  • Embeddings: using gemini's Embedding model, AlterEcho is able to create a vector database to efficiently store and read needed information.

2. Native Function Calling

The chatbot isn't just text. It uses Gemini Function Calling to autonomously interact with the world.

  • Tools: We define tools like generate_or_edit_image. Users don't need to type special commands. If you say "Send me a selfie" or "Make that picture darker", Gemini understands the intent ("Is this a tool call?") and executes the function naturally.

3. Multimodal Vision

AlterEcho can see.

  • Visual Understanding: You can upload images to the chat, and Gemini will analyze them and react in character (e.g., "Omg that cat is so cute!!" or "Where did you buy those shoes??").
  • Image Editing: It maintains an "Image History", allowing it to reference, understand, and even edit previous images in the conversation multimodally.

4. Low-Latency Streaming

For the voice feature (StreamChat), speed is critical.

  • Pipeline: As soon as Gemini generates the first few words, they are cleaned and piped directly into the WaveSpeed TTS engine. This results in a near-instant conversational flow that feels like a real phone call.

Tech Stack

  • Frontend: React 19, TailwindCSS v4, Radix UI, Framer Motion
  • Backend: Python 3.10+, Flask
  • AI: Google Gemini (Logic/Text/Vision/Image/Embeddings), WaveSpeed MiniMax (Voice)

All AI Models used in AlterEcho

  • Chatbot models : gemini-flash-latest, gemini-3-flash-preview, gemini-flash-lite-latest
  • Preprocessing models : gemini-3-pro-preview, gemini-2.5-pro, gemini-flash-latest, gemini-3-flash-preview, gemini-embedding-001
  • Imagen : gemini-2.5-flash-image, gemini-3-pro-image-preview
  • Voice cloning : WaveSpeed MiniMax 2.6 Turbo

Built With

Share this project:

Updates