spelling-bee-tts 0.4.2

GTK4 spelling quiz app with text-to-speech and AI generated example sentences.

Spelling Bee TTS

Spelling Bee TTS is a spelling testing app I wrote for my kids. It uses natural text-to-speech for the prompt and a local large language model to generate definitions and example sentences. It has a built in corpus of ~31,000 words.

The app tracks the user's progress and automatically adjusts the word difficulty level based on how well they're doing.

Install

System dependencies (Ubuntu):

sudo apt install python3-gi gir1.2-gtk-4.0 mpv

System dependencies (macOS with Homebrew):

brew install python gtk4 pygobject3 gobject-introspection mpv

Install with pip:

pip install spelling-bee-tts

Run

On Linux pip creates a application shortcut. Hit your launcher and search for 'spell'. Or to run from the command line:

python3 -m spellingbee

Notes

• edge-tts requires network access.
• You can override the voice with EDGE_TTS_VOICE (default: en-US-AriaNeural).
• Offline sentence generation uses llama-cpp-python with a local GGUF model.
  • By default, the app prompts to download Qwen3-4B-Instruct-2507 Q8_0 into the Hugging Face cache on first use.
  • Override with LLM_REPO_ID or LLM_MODEL_PATH.
  • Optional tuning: LLM_N_CTX, LLM_THREADS, LLM_N_BATCH, LLM_TEMPERATURE, LLM_TOP_P.