SpeechGPT

Inspired by ChatGPT, but via verbal communication

Development Status

FIRST DRAFT FINISHED

Installations

First, we need to install portaudio from Homebrew. This package is required for Python library pyaudio:

$ brew install portaudio

Then, we install packages in requirements.txt:

$ pip3 install -r requirements.txt

NOTE: If errors related to #include "portaudio.h" occurred when running pip install pyaudio, like this:

#include "portaudio.h"

         ^
1 error generated.
error: command '/usr/bin/clang' failed with exit status 1

Then we need to run the command separately in the Terminal below:

$ CFLAGS="-I/opt/homebrew/include -L/opt/homebrew/lib" python3 -m pip install pyaudio

We need to install FLAC in order to recognize speech with the command below:

$ sudo port install flac

If you have not installed Macports, please download Macports from this link.

Then, we run:

$ python3 src/main.py

NOTE: If you face the KeyError: 'VoiceAge' problem, please go to the nsss.py in your folder (the full error message should display the full path), and remove the attr['VoiceAge'] line code.

Before running this project, ensure that you have retrieve an OpenAI API key (you can sign up for an OpenAI account and create an API key here). Then, create an environment variable as below:

export OPENAI_API_KEY=<YOUR-OPENAI-API-KEY>

Now, you are ready to use SpeechGPT!

Instructions of Use

Run

python3 src/main.py

to kickstart SpeechGPT to listen.

Then, say "Hello!", "Hi!", "Hi there!", "Hello there!" when SpeechGPT is listening.

After SpeechGPT recognizes your words, it will be activated and greet.

Starting from now, you can simple talk to SpeechGPT as much as you want.

To deactivate SpeechGPT, please say "Stop!".

Dedications

I dedicate this hard work commitment to my mom, whom has sacrificed everything for my better future. I hope I have made you proud!