EchoMate

Record Your Voice

Speak into your microphone. The system will automatically detect when you start and stop speaking.

Transcription

No transcription available

Settings

Configure your speech-to-speech settings below, then click "Ready to Start" when you're ready to begin the conversation.

LLM Provider:

Localhost Server Web-based LLM (SmolLM2-1.7B-Instruct)

Chat Inference Server URL:

AI Character Card (v2):

No character card loaded

System Prompt:

Voice:

Character Information

View and edit character details. Upload a character card in the Settings tab to populate these fields.

No Avatar

Name:

Personality:

Description:

First Message:

Scenario:

System Prompt:

Creator:

Tags:

Credits

This project uses the following open source projects:

Speech2SpeechVAD

Main project source code repository

https://github.com/rhulha/Speech2SpeechVAD

Silero-VAD

Voice activity detection model

https://github.com/snakers4/silero-vad
Voice activity detector (VAD) for the browser by ricky0123

Moonshine

Speech recognition model by Useful Sensors

https://github.com/usefulsensors/moonshine

Kokoro

Text-to-speech synthesis engine

https://github.com/hexgrad/kokoro

Transformers.js

Machine learning library for running Transformers models in the browser

https://github.com/huggingface/transformers.js

SmolLM2-1.7B-Instruct

Language model by Hugging Face for instruction-following tasks

HuggingFaceTB/SmolLM2-1.7B-Instruct
Reddit discussion