hermes-breeze-tts

Breeze Blue text-to-speech as a Hermes TTS provider.

Install

Catalog (once listed):

hermes plugins install breeze-tts

Or by hand — the plugin package is the breeze-tts/ directory:

git clone https://github.com/MDsniper/hermes-breeze-tts
cp -R hermes-breeze-tts/breeze-tts ~/.hermes/plugins/breeze-tts

Then in ~/.hermes/config.yaml:

plugins:
  enabled: [breeze-tts]
tts:
  provider: breeze-blue
  voice: voc_...                     # Breeze voice id — required, there is no default
  model: breeze-tts-2                # optional

and the key:

echo 'BREEZE_API_KEY=brz_...' >> ~/.hermes/.env

hermes tools lists it as Breeze Blue and prompts for the key.

Notes

tts.voice is mandatory: Breeze has no built-in voice, every id is account-scoped. Voice ids, cloned voices and design previews come from the Breeze console.
Formats: mp3 (default), wav, flac, opus, aac, pcm. Opus is what Telegram voice bubbles need.
Direct HTTP, no breeze-blue SDK dependency: the API is ElevenLabs-shaped — POST /v1/text-to-speech/{voice_id}?output_format=mp3, key in xi-api-key.
BREEZE_BASE_URL overrides the endpoint for staging.
Not implemented: realtime/WebSocket streaming (streams_pcm), voice listing, async jobs.

Test

uv venv .venv && uv pip install pytest httpx
PYTHONPATH=~/path/to/hermes-agent .venv/bin/python -m pytest tests -q

breeze-tts

What it adds

Environment variables it needs 1

README