example.toml

[screenshot]
# The GPU ID device index. 1st GPU = 0, 2nd GPU = 1, etc.
GPU_IDX = 0

# The display monitor ID device index. 1st monitor = 0, 2nd monitor = 1, etc.
MONITOR_IDX = 0

# Sets resolution mode for resizing images prior to sending to LLM model while preserving aspect ratio.
# NOTE: These default are meant for OpenAI but work well for most other providers.
#   If 'false', resize screenshots such that height is 768.
#     See https://platform.openai.com/docs/guides/vision#calculating-costs for more info
#   If 'true', resize screenshots such that width is 512.
LOW_RESOLUTION_MODE = false

[chatbot]
# Select company provider for LLM chatbot model.
MODEL_PROVIDER = 'openai'

# Select LLM chat model name from company provider.
MODEL_NAME = 'gpt-4o'

# Keep at most 'n' back-and-forth interactions between human and bot. Set to 0 to always discard history.
KEEP_BACK_FORTH_HISTORY = 0

[prompts]
# The JSON file path for the list of prompts to load from.
PROMPTS_LIST_PATH = 'prompts/shounic.json'

# The key name of the specific prompt config from the aforementioned JSON file path.
PROMPT_CONFIG_NAME = 'normal'

[text_to_speech]
# Can specify the following based on which TTS service you want. There is a fallback mode to use Window TTS
# in case using the main TTS service fails for any reason.
# - 'online-ai' for ElevenLabs
# - 'offline-ai' for Coqui TTS (XTTS v2)
# - 'windows' for Microsoft's default TTS
MAIN_TTS_SERVICE = 'online-ai'

[text_to_speech.offline_ai]
# Glob path to list of example audio samples for use with voice cloning. Refer to https://pymotw.com/3/glob/ for examples.
VOICE_SAMPLES_PATH = 'scripts/example-data/voice.wav'

[text_to_speech.online_ai]
# Internal voice model ID for ElevenLabs, can either be from pre-existing models or custom-made cloned models.
ONLINE_VOICE_MODEL = 'example-voice-model'