GLaDOS Chat Library — Usage Guide

Talk with GLaDOS (Portal 2 style) via Ollama + Piper TTS + Whisper STT, either in your terminal or in Python scripts.

Features

• Interactive Chatting
  • One-shot answers or sentence-by-sentence streaming.
• Text-to-Speech (TTS)
  • Uses Piper with the GLaDOS voice model.
• Speech-to-Text (STT)
  • Uses Whisper to transcribe microphone input.
• Audio Device Control
  • Choose and persist your preferred output device.
• CLI with colors
  • Clean, color-coded output for user and GLaDOS lines.
• Sarcasm detection
  • Optional helper to flag user sarcasm.

Command Line Usage

Once installed, you have access to the glados-cli command.

1. Select your audio output device

glados-cli audiosource

• Lists all available audio devices.
• Prompts you to choose one.
• Saves your selection for future sessions in ~/.glados-cli.json.

2. Start a chat session

glados-cli start

Plain text chat (no TTS/STT).
Type exit or quit to leave.

3. Start with TTS

glados-cli start --tts

Replies are spoken in GLaDOS’s voice.

4. Start with Streaming

glados-cli start --stream

Replies print sentence-by-sentence as they’re generated.

5. Combine streaming + TTS

glados-cli start --stream --tts

Replies stream sentence-by-sentence and each is spoken aloud.

6. Override audio device (session only)

glados-cli start --tts --audiosource 27

Plays on device index 27 (from the device list).
Does not affect your saved config.

7. Enable Speech-to-Text

glados-cli start --stt

If integrated with a mic, transcriptions are fed into the chat.

Python Usage

Embed GLaDOS directly in your own scripts:

from glados_chat_lib import GladosChat

# Create a chat instance
chat = GladosChat(enable_tts=True, enable_stt=False, model="llama2:7b")

# One-shot conversation
reply = chat.ask("Do you hate humans?")
print("GLaDOS:", reply)
chat.speak(reply)

# Streaming mode
for piece in chat.ask("Tell me about science.", stream=True):
    print(piece, end="", flush=True)

Audio Devices

• Run glados-cli audiosource to choose and persist your output.
• To override without saving:

  glados-cli start --audiosource <index>
  

• To reset to default, delete ~/.glados-cli.json.

Quick Examples

Hear GLaDOS out loud in two steps:

glados-cli audiosource
glados-cli start --stream --tts

Embed in Python (minimal TTS only):

from glados_chat_lib import GladosChat

chat = GladosChat(enable_tts=True)
print("GLaDOS:", chat.ask("Sing me a song about testing."))