sttui 0.1.0

Linux-first speech-to-text terminal UI

sttui

No browser. No Web UI. Just fast speech-to-text in your terminal.

Features

• Records audio with pw-record.
• Shows an interactive Textual TUI with record/transcribe states.
• Sends WAV audio to OpenRouter as input_audio.
• Saves .wav and .txt outputs in a timestamped naming scheme.
• Supports --stdout mode for script-friendly output.

Requirements

• Linux with PipeWire (pw-record)
• Python 3.11+
• OpenRouter API key
• Optional clipboard tools: wl-copy or xclip

Install

# Recommended: isolated install for CLI tools
pipx install sttui

# Verify install
sttui --help

Alternative with pip:

python -m pip install --user sttui
sttui --help

Config

Create ~/.config/sttui/config.toml:

[openrouter]
api_key = "or-..."

[transcription]
model = "google/gemini-2.5-flash"
prompt = "Please transcribe this audio file."
max_seconds = 600

Commands

# Show CLI help
uv run sttui --help

# Start interactive dictation TUI
uv run sttui

# TUI + write transcript to stdout on Enter
uv run sttui --stdout

# Override model and recording cap for this run
uv run sttui --model google/gemini-2.5-flash --max-seconds 120

# Use a custom config file
uv run sttui --config ~/.config/sttui/config.toml

Keybindings:

• space / r: toggle record/stop
• s: stop recording
• c: copy transcript
• Enter: new cycle (or stdout confirm in --stdout mode)
• q: quit

CLI flags:

• --stdout
• --model <name>
• --max-seconds <int>
• --debug
• --config <path>

By default, recordings are stored in ~/.local/share/sttui/recordings/.

Development

uv sync
uv run sttui --help
uv run pytest