Teacher.ai

This project provides functionality to record audio until silence is detected, process the audio using a speech-to-text API, and generate a response with synthesized speech. The application is designed to run on a Jetson device with ALSA and Flask setup.

Table of Contents

• Prerequisites
• Installation
• Configuration
• Usage
• Troubleshooting

---

Prerequisites

Before you start, ensure the following dependencies are installed on your system:

1. Operating System: Linux-based system (e.g., Ubuntu) on Jetson devices.
2. Python: Python 3.8 or higher installed.
3. Packages:
   • Flask
   • Flask-SocketIO
   • pyaudio
   • requests
   • pydub
4. ALSA Utilities:
   • alsa-utils
   • arecord

Install ALSA utilities using:

sudo apt update
sudo apt install alsa-utils

5. Microphone and Speaker: Ensure your microphone and speaker are properly connected and recognized by the system.

---

Installation

1. Clone the repository:

git clone https://github.com/dextboy/treehacks.git
cd treehacks/backend

2. Set up a Python virtual environment (recommended):

python3 -m venv venv
source venv/bin/activate

3. Install Python dependencies:

pip install -r requirements.txt

4. Install additional required libraries:

   • pyaudio may require the portaudio library. Install it using:

   sudo apt install libportaudio2 libportaudiocpp0 portaudio19-dev

5. Configure ALSA to use the correct audio device:

   • Edit the ~/.asoundrc file:

     pcm.!default {
         type hw
         card 2
         device 0
     }
     
     ctl.!default {
         type hw
         card 2
     }

     Replace card and device with the appropriate values for your system. Use arecord -l to identify them.

---

Configuration

1. Set Up Flask Application:

   • Run the application by navigating to the project directory and starting the Flask server:

     python app.py

2. ALSA Setup:

   • Restart ALSA or reload its configuration:

     sudo alsactl init

---

Usage

1. Start Recording Until Silence: Run the application, and it will start recording audio until silence is detected. The recorded file will be saved as output.wav.

2. Process Recorded Audio: The output.wav file will automatically be sent to the /process/speech API endpoint for transcription and speech synthesis.

3. Play Synthesized Speech: The synthesized audio file will be played on the device.

---

Troubleshooting

1. ALSA Errors:

   • If you encounter errors like Unknown PCM or Cannot open device, verify your .asoundrc configuration and check connected audio devices using:

     aplay -l
     arecord -l

2. PyAudio Installation:

   • If pyaudio fails to install, ensure the portaudio library is installed:

     sudo apt install libportaudio2 libportaudiocpp0 portaudio19-dev

3. Permissions Issues:

   • Ensure you have the required permissions to access audio devices:

     sudo usermod -aG audio $USER

4. Device Not Recognized:

   • Check the microphone and speaker connection. Restart ALSA with:

     sudo alsactl init

---