TUI for OpenAI

Overview

TUI for OpenAI is a lightweight terminal-based utility that allows users to send prompts to OpenAI's API via text or voice input. The tool consists of two Bash scripts:

1. discuss.sh – Captures voice input, transcribes it using Whisper, and sends it to OpenAI's API.
2. introduce.sh – Sends a written prompt to OpenAI's API for a faster, more efficient interaction.

Both scripts require an OpenAI API key, which must be stored as an environment variable.

Installation & Setup

Prerequisites

Ensure you have the following installed on your system:

• Bash (for running the scripts)
• Whisper (for speech-to-text transcription)
• jq (for handling JSON responses)
• curl (for making API requests)

Setup

1. Clone the repository:

   git clone https://github.com/your-repo/tui-for-openai.git
   cd tui-for-openai

2. Add your OpenAI API key to your shell environment:

   export OPENAI_API_KEY="your-api-key-here"

   To make this persistent, add the above line to your .bashrc or .zshrc file.

Usage

1. Using discuss.sh (Voice Input to OpenAI)

This script records an audio prompt, transcribes it, and sends it to OpenAI.

./discuss.sh

Process:

• Records audio (prompt.wav)
• Converts speech to text (prompt.txt)
• Sends text to OpenAI API
• Receives response and saves it in output.txt

2. Using introduce.sh (Text Input to OpenAI)

This script allows users to send a written prompt stored in a text file.

./introduce.sh prompt.txt

Process:

• Reads text from prompt.txt
• Sends it to OpenAI API
• Saves the response in output.txt

Code Snippets

discuss.sh (Voice to Text to OpenAI)

#!/bin/bash
AUDIO_FILE="prompt.wav"
OUTPUT_FILE="prompt.txt"
RESPONSE_FILE="response.json"
FINAL_OUTPUT="output.txt"

echo "Recording audio as $AUDIO_FILE... Press Ctrl+C to stop."
arecord -f cd -t wav -d 0 -q "$AUDIO_FILE"

whisper "$AUDIO_FILE" --model tiny --output_format txt --output_dir . > /dev/null 2>&1
echo "Transcription complete and saved to $OUTPUT_FILE."

PROMPT=$(jq -Rs . "$OUTPUT_FILE")
echo "Sending prompt to OpenAI API..."

JSON_PAYLOAD=$(jq -n \
	--arg model "gpt-4o-mini" \
	--argjson messages "[{\"role\": \"user\", \"content\": $PROMPT}]" \
	'{model: $model, messages: $messages}')

HTTP_STATUS=$(curl -s -w "%{http_code}" -o $RESPONSE_FILE https://api.openai.com/v1/chat/completions \
	-H "Authorization: Bearer $OPENAI_API_KEY" \
	-H "Content-Type: application/json" \
	-d "$JSON_PAYLOAD") 

if [ -s $RESPONSE_FILE ]; then
	if jq -e . >/dev/null 2>&1 < $RESPONSE_FILE; then
		cat $RESPONSE_FILE | jq -r '.choices[0].message.content' > $FINAL_OUTPUT
	else
		echo "Invalid JSON response:"
		cat $RESPONSE_FILE
	fi 
else
	echo "Response file is empty or contains an error."
fi

echo "Process complete and saved to $FINAL_OUTPUT."

introduce.sh (Text to OpenAI)

#!/bin/bash
INPUT_FILE="$1"
RESPONSE_FILE="response.json"
FINAL_OUTPUT="output.txt"

PROMPT=$(jq -Rs . "$INPUT_FILE")

echo "Sending prompt to OpenAI API..."
JSON_PAYLOAD=$(jq -n \
	--arg model "gpt-4o-mini" \
	--argjson messages "[{\"role\": \"user\", \"content\": $PROMPT}]" \
	'{model: $model, messages: $messages}')

HTTP_STATUS=$(curl -s -w "%{http_code}" -o $RESPONSE_FILE https://api.openai.com/v1/chat/completions \
	-H "Authorization: Bearer $OPENAI_API_KEY" \
	-H "Content-Type: application/json" \
	-d "$JSON_PAYLOAD")

if [ -s $RESPONSE_FILE ]; then
	if jq -e . >/dev/null 2>&1 <$RESPONSE_FILE; then
		cat $RESPONSE_FILE | jq -r '.choices[0].message.content' > $FINAL_OUTPUT
	else
		echo "Invalid JSON response:"
		cat $RESPONSE_FILE
	fi
else
	echo "Response file is empty or contains an error."
fi

echo "Process complete and saved to $FINAL_OUTPUT."

File Management

During usage, the following files will be created in the working directory:

These files will be automatically overwritten with each new request, so manual cleanup is not required.

Future Improvements

• Add support for multi-turn conversations using a persistent chat history.
• Improve error handling for invalid API responses.
• Integrate OpenAI’s function calling to interact with external tools.

License

This project is open-source.

Conclusion

TUI for OpenAI provides a simple and efficient way to interact with OpenAI's API using either voice or text input. It’s designed to be lightweight, fast, and flexible for users who prefer a terminal-based workflow.