blob: 2378079b5d28ca922c7211e3fe39fb311c2b9eb3 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
post_install() {
cat <<'MSG'
==> voice-agent installed.
First run: voice-agent
(bootstraps ~/.local/share/voice-agent + a uv venv; downloads the
Python deps + STT/TTS models on first launch — multi-GB, needs network)
Config: edit ~/.local/share/voice-agent/.env — pick a brain (it's pluggable):
default local Anthropic brain → set ANTHROPIC_API_KEY
fully local → LLM_PROVIDER=ollama (no cloud key)
external brain → BRAIN=gabagent (voice shell needs no LLM key)
Brains: voice-agent (brain per .env BRAIN=)
voice-agent gab (gabagent brain, if installed)
voice-agent local (local LLM: Ollama/Anthropic/OpenAI per .env)
voice-agent debug (gabagent + brain-side debug log)
Update: after a package upgrade, run voice-agent --update
(re-syncs code; preserves your .env / wakewords / logs / models / venv)
Audio: needs a working PipeWire/PulseAudio setup + a microphone.
Wake word: the speaker-specific model is NOT shipped — open-mic works without it;
train your own per the repo's wake-train docs.
MSG
}
post_upgrade() {
cat <<'MSG'
==> voice-agent upgraded. Run voice-agent --update to sync the new code into your
working copy (~/.local/share/voice-agent), preserving your config + data.
MSG
}
|