README.md

metadata

title: Good Morning
emoji: ☀️
colorFrom: yellow
colorTo: red
sdk: static
pinned: false
short_description: First app for the new Reachy Mini, by Clem Delangue
tags:
  - reachy_mini
  - reachy_mini_python_app

Good Morning — a Reachy Mini app

First! The first Reachy Mini app built for the new version of Reachy Mini, by Clem Delangue, co-founder & CEO of Hugging Face.

Press a button → Reachy Mini says:

"Good morning. Today is a good day to help more people become AI builders, not just AI users thanks to open-source!"

The button lives at http://localhost:7860/ whenever the app is running on your robot. The app also auto-opens that page in your default browser when it starts.

Install

From the project root:

# Same Python env you use for the reachy-mini SDK (3.10–3.12).
pip install -e .

This installs the reachy_mini_apps entry point good_morning, so the app shows up in the Reachy Mini dashboard's app list.

Run

Two options:

1) Through the Reachy Mini app system (preferred — this is what makes it "an app"). Install from the Reachy Mini dashboard's Discover Apps panel, hit Start on the good_morning tile, and the app will pop your default browser open at http://localhost:7860/. Click the orange button.

2) Standalone, for quick testing:

python -m good_morning_app.main

Then open http://localhost:7860/ and click the button.

How it works

The first time the app starts, pyttsx3 synthesises the phrase once and caches it as good_morning.<hash>.wav next to the code (the hash is derived from the phrase, so changing PHRASE auto-invalidates the cache). On every press we:

1. Load the cached WAV → mono float32, resampled to 16 kHz, shape (N, 1).
2. Push it to the speaker with reachy_mini.media.push_audio_sample(samples) (the canonical TTS path per the SDK docs — there is no .speak() helper).
3. Sleep for the audio's duration since push_audio_sample is non-blocking.

A small thread polls the local FastAPI's /healthz endpoint until it answers 200, then opens http://localhost:7860/ in your default browser.

Tweaking

Open src/good_morning_app/main.py:

• PHRASE — change what it says (audio cache is regenerated automatically).
• WEB_PORT — where the button page lives.

To swap the TTS engine (e.g. for Piper or an online voice), replace _synthesize_phrase() — anything that produces a WAV will work; the loader handles mono/stereo and resampling.

Project layout

reachy-mini-good-morning/
├── pyproject.toml          # registers the reachy_mini_apps entry point
├── README.md               # this file (HF Space frontmatter at the top)
└── src/
    └── good_morning_app/
        ├── __init__.py
        └── main.py         # GoodMorningApp(ReachyMiniApp)

References

• Reachy Mini SDK: https://github.com/pollen-robotics/reachy_mini
• Audio API (push_audio_sample): https://github.com/pollen-robotics/reachy_mini/blob/main/docs/source/SDK/python-sdk.md
• Reference app (entry-point shape, ReachyMiniApp subclass): https://github.com/pollen-robotics/reachy_mini_conversation_app