SPF — Smart Car AI | Multilingual Voice Assistant

Languages Supported

Language	Voice Commands	Responses
English	✅	✅
Hindi	✅ (romanised)	✅
Telugu	✅ (romanised)	✅
Spanish	✅	✅

Project Structure

SPF_Final/
├── api.py            ← Flask server (run this)
├── language.py       ← Language detection + multilingual responses (NEW)
├── intent_model.py   ← LSTM trained on EN+HI+TE+ES phrases
├── entity.py         ← Multilingual entity extractor
├── action.py
├── context.py
├── preprocess.py
├── speech.py
├── wakeword.py
├── requirements.txt
└── templates/
    └── dashboard.html ← Full UI with 4-language mic support

Install

pip install flask flask-cors speechrecognition pyttsx3 nltk tensorflow pyaudio

If NLTK errors on first run:

import nltk; nltk.download('punkt'); nltk.download('punkt_tab')

Run

python api.py

Open: http://localhost:5000 (Chrome or Edge)

Example Voice Commands

English

• "Set temperature to 24 degrees"
• "Open the window"
• "Play music"
• "Navigate to home"
• "Call mom"

Hindi

• "Temperature 24 karo"
• "Khidki kholo"
• "Gaana bajao"
• "Ghar chalo"
• "Maa ko call karo"

Telugu

• "Temperature 24 cheyyi"
• "Kiddiki teruvu"
• "Paata veyyi"
• "Illu ki vellu"
• "Amma ki call cheyyi"

Spanish

• "Pon la temperatura a 24"
• "Abre la ventana"
• "Pon musica"
• "Navega a casa"
• "Llama a mama"

How Language Detection Works

1. Browser's Web Speech API runs 4 recognizers in parallel (EN/HI/TE/ES)
2. First one to get a result wins
3. Flask's language.py scores keyword matches to confirm language
4. Response is generated in the detected language
5. Memory panel shows detected language with flag