CleanDictate

Intelligent Dictation Engine with real-time speech-to-text, filler removal, and dual tone transformation (Professional & Casual).

Requirements

• Python 3.10+
• Intel i5+ / AMD5+
• Windows 10/11

Installation

# Clone the repository
git clone https://github.com/SharadhNaidu/CleanDictate.git
cd CleanDictate

# Create virtual environment
python -m venv venv
venv\Scripts\activate

# Install PyTorch with CUDA
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

# Install dependencies
pip install faster-whisper transformers spacy pyaudio pynput scipy

# Download spaCy model
python -m spacy download en_core_web_sm

Run

python cleandictate.py

Hotkeys

Key	Action
F9	Start Recording
F10	Stop Recording
F11	Copy Output
ESC	Quit

Modes

• Live: Real-time VAD, instant typing
• Document: Record all, process on stop

Tones

• neutral - No style change
• formal - Professional business tone
• casual - Friendly conversational tone
• concise - Brief and direct
• dual - Both formal & casual output