bopi

KO | EN

gitlite — search

#python #hacktoberfest #react #ai #typescript #go #machine-learning #docker #nodejs #kubernetes #vue #pytorch #openai #mcp

(★ 76)

Your DIY Backpack Pet

bopi 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Bopi: Your DIY Backpack Pet

[![X](https://img.shields.io/badge/X-@pham__blnh-000000?style=flat&logo=x&logoColor=white)](https://x.com/pham_blnh/status/2037897444286169505?s=20)

A voice agent companion running on [SenseCAP Watcher](https://www.seeedstudio.com/SenseCAP-Watcher-W1-A-p-5979.html), powered by [LiveKit](https://livekit.io) and [Dasai](https://dasai.com.au) animations.

Talk to it, and it reacts with expressions on screen.

**Quick links**:

- [**Build It Yourself**](#what-you-need)
- [**How It Works**](#how-it-works)
- [**Offline Version**](https://github.com/pham-tuan-binh/watcher-mochi)

## What You Need

- SenseCAP Watcher: [Buy here - $69 - Coupon: 5EB420ZS](https://www.seeedstudio.com/SenseCAP-Watcher-W1-A-p-5979.html?sensecap_affiliate=3gToNR2&referring_service=link)
- A microSD card (any size, FAT32 formatted)
- A USB-C cable
- A computer with [ESP-IDF](https://docs.espressif.com/projects/esp-idf/en/stable/esp32s3/get-started/) v5.4+ installed
- A [LiveKit Cloud](https://cloud.livekit.io) project (or self-hosted server)
- The [LiveKit CLI](https://docs.livekit.io/home/cli/cli-setup/) (`lk`)

❤️ **If you want to buy a SenseCAP Watcher, consider using the link or coupon above.** It's an affiliate link so I'll get a small percentage of your order as appreciation ^^

## Step 1: Prepare the SD Card

1. Format your microSD card as **FAT32**
2. Copy all the `.gif` files from the `sd_content/` folder onto the root of the SD card
3. Insert the SD card into your Watcher

There are 63 animations included plus a `blank.gif` that shows between expressions. You can add your own GIFs too - just drop any `.gif` file onto the SD card root.

## Step 2: Configure

Settings can be set through `idf.py menuconfig` or added directly to `sdkconfig`:

### Credentials

**Option A** - Use a LiveKit Sandbox for quick setup. Create one from your [Cloud Project](https://cloud.livekit.io/projects/p_/sandbox):

```ini
CONFIG_LK_BOPI_USE_SANDBOX=y
CONFIG_LK_BOPI_SANDBOX_ID="my-project-xxxxxx"
```

**Option B** - Use a pre-generated token and server URL:

```ini
CONFIG_LK_BOPI_USE_PREGENERATED=y
CONFIG_LK_BOPI_SERVER_URL="wss://your-project.livekit.cloud"
CONFIG_LK_BOPI_TOKEN="<your-token>"
```

To generate a token with the LiveKit CLI:

```sh
lk token create --project \
    --join --room bopi --agent "bopi-agent" --identity bopi \
    --valid-for 24h
```

This creates a token that joins the room `bopi`, dispatches the agent named `bopi-agent`, and identifies the Watcher as `bopi`.

### Network

WiFi:

```ini
CONFIG_LK_EXAMPLE_USE_WIFI=y
CONFIG_LK_EXAMPLE_WIFI_SSID="<your SSID>"
CONFIG_LK_EXAMPLE_WIFI_PASSWORD="<your password>"
```

> Note: WiFi and Ethernet settings come from the LiveKit `example_utils` component, so they still use the `LK_EXAMPLE_` prefix.

## Step 3: Run the Agent

The `agent/` folder contains a LiveKit Agents server that powers Bopi's voice and expressions. You need [uv](https://docs.astral.sh/uv/) and a `.env.local` file with your API keys.

Generate your `.env.local` credentials:

```sh
cd agent
lk app env -w
```

Then start the agent:

```sh
cd agent
uv sync
uv run bopi-agent dev
```

The agent will connect to your LiveKit project and wait for the Watcher to join the room.

## Step 4: Build and Flash the Firmware

```sh
idf.py build
idf.py flash monitor
```

Press `Ctrl+]` to exit the monitor.

## How It Works

Bopi is built on top of [LiveKit's ESP32 SDK](https://github.com/livekit/client-sdk-esp32), which handles real-time audio streaming over WebRTC directly on the microcontroller. The Watcher connects to a LiveKit room, publishes its microphone audio, and subscribes to audio from the agent - all running on the ESP32-S3.

On the server side, a [LiveKit Agent](https://docs.livekit.io/agents/) (`agent/`) listens to the room. It uses speech-to-text, an LLM, and text-to-speech to hold a conversation. The agent's responses are streamed back to the Watcher as audio, and its transcription is sent as a data stream so the Watcher can react to individual words in real time.

Here's what to expect when using it:

1. **Tap the screen** - the Watcher connects to the LiveKit room and the agent is dispatched
2. **Start talking** - the agent hears you, thinks, and talks back through the speaker
3. **Watch the expressions** - as the agent speaks, matching animations play on screen
4. **Tap again** to mute/unmute the mic
5. **Long-press** to disconnect and restart

The expressions work through the agent's transcription data stream (`lk.transcription`). As the agent speaks, each word is checked against the GIF filenames on the SD card. If a word matches, that animation plays on screen. For example, if the agent says "happy", `happy.gif` plays. If no GIF matches, the word is shown as text instead. The matching is case-insensitive and strips trailing punctuation, so "Happy!" still triggers `happy.gif`. Only single words are matched to keep things snappy.

|                                |                                    |                                |                                  |                                        |
| :----------------------------: | :--------------------------------: | :----------------------------: | :------------------------------: | :------------------------------------: |
| ![happy](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/happy.gif) | ![dancing](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/dancing.gif) |  ![love](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/love.gif)  | ![sleepy](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/sleepy.gif) | ![surprised](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/surprised.gif) |
|             happy              |              dancing               |              love              |              sleepy              |               surprised                |
| ![devil](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/devil.gif) | ![sparkle](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/sparkle.gif) | ![sushi](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/sushi.gif) |   ![rain](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/rain.gif)   |      ![wink](https://raw.githubusercontent.com/pham-tuan-binh/bopi/main/sd_content/wink.gif)      |
|             devil              |              sparkle               |             sushi              |               rain               |                  wink                  |

Between expressions, a `blank.gif` idle animation is shown.

## License

The firmware source code is licensed under the [Apache License 2.0](LICENSE).

The GIF animations in `sd_content/` are property of [Dasai](https://dasai.com.au) and are included here for personal use with the Bopi project. All rights to the animations belong to Dasai.

bopi

bopi 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

bopi

bopi 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README