PersonalAssistant

KO | EN

gitlite — search

#javascript #react #typescript #llm #golang #machine-learning #deep-learning #linux #docker #nodejs #cli #vue #pytorch #security

PersonalAssistant

(★ 26)

AI personal assistant written in Python. Voice Assistant with Real-Time Speech Recognition and Response on a local computer.

PersonalAssistant 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# 🎙️ Voice Assistant with Real-Time Speech Recognition and Response

This is a real-time voice assistant built in Python using [OpenAI Whisper](https://github.com/openai/whisper), [WebRTC Voice Activity Detection (VAD)](https://github.com/wiseman/py-webrtcvad), and a local LLM server for generating responses. The assistant listens for speech when you press the spacebar, transcribes it using Whisper, sends the transcribed text to a chatbot endpoint, and optionally speaks the response aloud using gTTS.

## Features

- Real-time microphone input and recording
- On-device Whisper speech-to-text transcription
- Voice activity detection (VAD) for intelligent segmentation
- Chatbot response via local LLM API
- Text-to-speech response using `gTTS_module`
- Terminal theming (light/dark)
- Multithreaded design for responsive interaction

## 📦 Dependencies

Install required libraries via pip:

```bash
pip install numpy sounddevice keyboard openai-whisper torch webrtcvad requests colorama
```

> You also need a working `gTTS_module.py` in the same directory with a function:  
> `text_to_speech_withEsc(text: str)` that plays audio and exits on `Esc`.

## 🛠️ Usage

1. **Run the script**:
   ```bash
   python pers_assist.py
   ```

2. **Press the `Space` key** to toggle voice recording.

3. **Speak** into your microphone. When you stop talking, the assistant transcribes your speech and responds.

4. The assistant **prints the response** in the terminal and **speaks it aloud**.

## 📡 LLM Chatbot Backend

You can use this assistant with a local language model by installing [LM Studio](https://lmstudio.ai), which provides an easy interface for running models locally.

For example, you can load and serve the model [`google/gemma-3-4b`] in LM Studio.

Make sure LM Studio is configured to serve an OpenAI-compatible API at:

```
http://localhost:1234/v1/chat/completions
```

This assistant expects a local language model server running at:

```
http://localhost:1234/v1/chat/completions
```

The request format sent is:
```json
{
  "messages": [{"role": "user", "content": "your question"}]
}
```

Make sure your backend server supports this structure (e.g. OpenAI-compatible).

## 🧠 Model Selection

Whisper models supported:
- `tiny`, `base`, `small`, `medium`, `large`, etc.
- `.en` suffix for English-only variants, etc.

## 🎨 Themes

Two terminal color themes are available:
- `"light"` (default)
- `"dark"`

You can switch by changing:
```python
THEME = THEMES["light"]
```

## 🧩 File Structure
```
├── pers_assist.py         # Main application
├── gTTS_module.py         # Text-to-speech utility
└── README.md              # This file
```
## 📝 Notes

- Pressing the spacebar toggles recording on/off.
- The assistant detects silence to trigger transcription and response generation.
- VAD settings are tuned for responsiveness and low false positives.

## 📄 License

General Public License. See `LICENSE` file for details.

---
Join and have fun!

PersonalAssistant

PersonalAssistant 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

PersonalAssistant

PersonalAssistant 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README