turing

A general-purpose AI assistant with voice IO support and custom system prompts powered by OpenAI. turing can receive it's input prompt from the microphone (leveraging OpenAI's STT model), from a file or from standard input. The input is then processed by the target model, which may be a GPT or one of the image generation models. The output is then played on the system speakers (leveraging OpenAI's TTS model), written to a file or simply output to the console. The model's response can also be copied to the clipboard.

From the beggining, turing was thought of as a tool to be used via keyboard shortcuts with preconfigured parameters, where the user writes his/her own custom system prompts and simply types a keyboard sequence, speaks to the microphone and hears the response spoken back and/or has it copied to the clipboard.

Features

• TTS (text-to-speech).
• STT (speech-to-text).
• Text generation (GPTs).
• Image generation.
• Setting custom system prompts for the GPT models.

Setup

To setup turing, inside src/, copy the example file config-example.py into config.py and add your organization ID and API key. Some of the options for the models are only configurable via this file, so they can be changed here.

Examples

$ turing -h
turing [options]
        -h, --help                  show this banner
        -v, --verbose               enable runtime messages
        -c, --clipboard             copy the response to the clipboard
        -l, --list-models           list the available models
        -i, --input [source]        source of the prompt (can be "stdin" (default), "audio" (microphone) or a filename)
        -o, --output [destination]  destination of the model's response (can be "stdout" (default), "audio" (speakers) or a filename)
        -m, --model [model]         set the OpenAI model to use (default: gpt-3.5-turbo)
        -s, --system [prompt]       set a system prompt for the GPT model
        -S, --system-file [file]    read a system prompt for the GPT model from [file]

        * Microphone input transcription (-i audio) and audio generation (-o audio) increase API usage.

• Transcribe the prompt from the microphone and listen to the default model's (gpt-3.5-turbo) response

turing -i audio -o audio

• Read the prompt from a file and listen to GPT-4's response

turing -vm gpt-4o -i ./prompt.txt -o audio

• Transcribe the prompt from the microphone and generate an image using Dall-E v3. Save the output to "image.png"

turing -vm dall-e-3 -i audio -o image.png

• Transcribe the microphone's input and copy to the clipboard

turing -vcm whisper-1 -i audio

• Convert your voice to the TTS model's voice and save it to voice.wav

turing -vm tts-1 -i audio -o voice.wav

System Prompts

Here are examples of some useful system prompts:

• Generate commands

You are an assistant specialized in providing CLI commands to perform various tasks. Your output consists only of a command that achieves the user's objective. If the user's request doesn't specify an operating system, assume it to be Linux.

• Generate source code

You are an assistant specialized in programming whose purpose is to provide code implementations. Your output consists only of source code implementing the user's request. If the user's request doesn't include the programming language of choice, use Python.