Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/ximilalaxiang/delive

System audio capture + multi-provider ASR + local-first AI review workspace. Floating live captions, 6 ASR backends, 60+ languages, AI summary/chat/mindmap, Open API, MCP server, and Agent Skill.
https://github.com/ximilalaxiang/delive

agent-skill ai asr captions desktop-app electron groq linux local-first mcp open-api realtime soniox speech-recognition speech-to-text stt tailwindcss transcription typescript windows

Last synced: about 1 month ago
JSON representation

System audio capture + multi-provider ASR + local-first AI review workspace. Floating live captions, 6 ASR backends, 60+ languages, AI summary/chat/mindmap, Open API, MCP server, and Agent Skill.

Awesome Lists containing this project

README 

          
DeLive Header






# DeLive



**System-Level Audio Capture | Cloud and Local ASR in One Desktop App**



English | [简体中文](./README_ZH.md) | [繁體中文](./README_TW.md) | [日本語](./README_JA.md)



[![Version](https://img.shields.io/github/v/release/XimilalaXiang/DeLive?label=Version&color=blue)](https://github.com/XimilalaXiang/DeLive/releases)

[![License](https://img.shields.io/github/license/XimilalaXiang/DeLive?label=License&color=green)](https://github.com/XimilalaXiang/DeLive/blob/main/LICENSE)

[![Platform](https://img.shields.io/badge/Windows-0078D6?logo=windows&logoColor=white)](https://github.com/XimilalaXiang/DeLive/releases)

[![Platform](https://img.shields.io/badge/macOS-000000?logo=apple&logoColor=white)](https://github.com/XimilalaXiang/DeLive/releases)

[![Platform](https://img.shields.io/badge/Linux-FCC624?logo=linux&logoColor=black)](https://github.com/XimilalaXiang/DeLive/releases)

[![Downloads](https://img.shields.io/github/downloads/XimilalaXiang/DeLive/total?label=Downloads&color=orange)](https://github.com/XimilalaXiang/DeLive/releases)

[![Stars](https://img.shields.io/github/stars/XimilalaXiang/DeLive?style=social)](https://github.com/XimilalaXiang/DeLive)



[Core Features](#-core-features) • [Quick Start](#-quick-start) • [Architecture](#-system-architecture) • [Providers](#-supported-asr-providers)






DeLive captures system audio output directly. If your computer can play the sound, DeLive can capture it, feed it into the ASR backend you choose, and keep the resulting transcript on your machine for review, export, and reuse.





DeLive Screenshot




## 🎯 Core Features



- **System-level audio capture** for browser video, live streams, meetings, courses, and any other playback source that exposes shareable system audio.

- **Cloud and local ASR backends** in one app: Soniox, Volcengine, OpenAI-compatible local services, and experimental `whisper.cpp`.

- **Provider-aware audio pipeline** that switches between `MediaRecorder` and PCM16 processing based on backend requirements.

- **Local model workflows** including service detection, installed-model discovery, optional Ollama one-click pull, and `whisper.cpp` binary/model import or download.

- **Floating caption overlay** with always-on-top transparent window, draggable mode, and style customization.

- **History, tags, search, and export** with TXT and SRT output.

- **Desktop integration** with tray behavior, global shortcut, auto-launch, update checks, and bilingual UI.



## 🏗️ System Architecture



```mermaid

graph TB

    subgraph "Desktop Shell"

        EM[Electron Main Process]

        CAP[Caption Window
Transparent Overlay]

        DESK[Tray / Shortcut / Auto Launch / Updater]

    end



    subgraph "Frontend"

        UI[React App + Zustand]

        CFG[Provider Config UI]

        HIS[History / Export / Backup]

    end



    subgraph "Capture Pipeline"

        SRC[Source Picker]

        GDM[getDisplayMedia]

        MR[MediaRecorder
WebM / Opus]

        AP[AudioProcessor
PCM16 16kHz]

    end



    subgraph "Provider Layer"

        REG[Provider Registry]

        SON[Soniox Provider]

        VOL[Volc Provider]

        LOA[Local OpenAI-compatible Provider]

        WCP[whisper.cpp Runtime Provider]

    end



    subgraph "Electron Services"

        PROXY[Embedded Volc Proxy
Express + WebSocket]

        RTM[Local Runtime Manager
IPC + Process Control]

    end



    subgraph "ASR Backends"

        SONIOX[Soniox Realtime API]

        VOLC[Volcengine API]

        OPENAI[Ollama / OpenAI-compatible ASR]

        WHISPER[whisper.cpp server]

    end



    subgraph "Persistence"

        STORE[Local Storage
settings / sessions / tags]

    end



    UI --> CFG

    UI --> HIS

    UI --> SRC

    SRC --> GDM

    GDM --> MR

    GDM --> AP



    UI --> REG

    REG --> SON

    REG --> VOL

    REG --> LOA

    REG --> WCP



    MR --> SON

    MR --> LOA

    AP --> VOL

    AP --> WCP



    SON --> SONIOX

    VOL --> PROXY

    PROXY --> VOLC

    LOA --> OPENAI

    WCP --> RTM

    RTM --> WHISPER



    UI --> STORE

    HIS --> STORE

    UI --> EM

    EM --> CAP

    EM --> PROXY

    EM --> RTM

    EM --> DESK



    style UI fill:#61dafb,color:#000

    style EM fill:#334155,color:#fff

    style CAP fill:#f472b6,color:#000

    style REG fill:#f59e0b,color:#000

    style PROXY fill:#10b981,color:#fff

    style RTM fill:#0f766e,color:#fff

    style OPENAI fill:#8b5cf6,color:#fff

    style WHISPER fill:#ef4444,color:#fff

```



### Architecture Overview



| Layer | Main Components | Notes |

|-------|-----------------|-------|

| Desktop shell | Electron main process, tray, updater, caption window | Owns native desktop behavior and IPC |

| Frontend | React, Zustand, provider config UI, history/export UI | Manages recording flow, settings, session state |

| Capture pipeline | `getDisplayMedia`, `MediaRecorder`, `AudioProcessor` | Picks encoding path based on provider capability |

| Provider layer | Registry + 4 provider implementations | Normalizes cloud and local ASR usage behind one interface |

| Electron services | Embedded Volc proxy, bundled runtime manager | Handles custom-header WebSocket proxying and local process lifecycle |

| Persistence | Browser local storage | Stores settings, transcript sessions, and tags locally |



## 🔌 Supported ASR Providers



| Provider | Type | Audio Path | Highlights |

|----------|------|------------|------------|

| **Soniox V4** | Cloud | `MediaRecorder` -> WebSocket | Token-level realtime transcription, multi-language |

| **Volcengine** | Cloud | PCM16 -> embedded proxy -> WebSocket | Chinese-optimized flow, proxy handles required headers |

| **Local OpenAI-compatible** | Local service | `MediaRecorder` -> `/v1/audio/transcriptions` | Works with Ollama or compatible gateways, model discovery and optional Ollama pull |

| **Local whisper.cpp** | Local runtime | PCM16 -> local `/inference` | Experimental; can import/download `whisper-server` binaries and local `.bin` / `.gguf` models |



## 🚀 Quick Start



### Prerequisites



- Node.js 18+

- Choose one backend path:

  - Soniox API key

  - Volcengine APP ID and Access Token

  - A local OpenAI-compatible ASR service that exposes `/v1/models` and `/v1/audio/transcriptions`

  - A `whisper.cpp` server binary plus a local model file, or let DeLive download/import them from the setup guide



### Installation



```bash

git clone https://github.com/XimilalaXiang/DeLive.git

cd DeLive

npm run install:all

```



### Development



```bash

npm run dev

```



`npm run dev` starts Vite and Electron together. The Volcengine proxy used by the desktop app is already embedded in `electron/main.ts`.



If you need the standalone proxy for debugging or non-Electron experiments, run:



```bash

npm run dev:server

```



### Build



```bash

npm run dist:win

npm run dist:mac

npm run dist:linux

```



Artifacts are written to `release/`.



### Optional: Stage `whisper.cpp` Into Packaged Builds



```bash

# Fetch an official release asset and stage it into local-runtimes/whisper_cpp/

npm run fetch:whisper-runtime -- --target win32



# Or stage your own binary explicitly

npm run stage:whisper-runtime -- --binary /path/to/whisper-server --target linux

```



If `local-runtimes/whisper_cpp/whisper-server(.exe)` exists at build time, `electron-builder` packages it as an extra resource. End users can still import or download binaries and models later from the UI.



## 📖 Usage



### Cloud Providers



1. Open settings and pick `Soniox V4` or `Volcengine`.

2. Enter the required credentials and run `Test Config`.

3. Click `Start Recording`.

4. Choose a screen or window and make sure audio sharing is enabled.

5. Watch partial and final transcripts update in the main window or caption overlay.



### Local OpenAI-compatible Services



1. Select `Local OpenAI-compatible`.

2. Fill in `Base URL` and `Model`.

3. Use the local setup guide to detect the service and list installed models.

4. If the detected service is Ollama, you can pull the selected model directly from the app.



### Local `whisper.cpp` Runtime



1. Select `Local whisper.cpp`.

2. Prepare a runtime binary by importing an existing `whisper-server` file or downloading a recommended official asset.

3. Prepare a model by selecting, importing, or downloading a local `.bin` or `.gguf` file.

4. Start the runtime or run `Test Config`.

5. Record normally; DeLive will launch and talk to the local runtime through Electron IPC.



### Captions, History, and Export



- Toggle the floating caption window and adjust font, colors, size, width, shadow, and position.

- Review saved sessions in the history panel, rename them, and organize them with tags.

- Export transcripts as TXT or SRT.

- Import or export all local data from the settings panel for backup or migration.



## 📁 Project Structure



```text

DeLive/

├── electron/                       # Electron main process and IPC bridge

│   ├── main.ts                     # Embedded proxy, runtime manager, tray, updater

│   └── preload.ts                  # Renderer-safe Electron API surface

├── frontend/

│   ├── caption.html                # Separate entry for the caption overlay window

│   ├── src/

│   │   ├── components/             # UI panels and setup guides

│   │   ├── hooks/                  # Recording and ASR orchestration

│   │   ├── providers/              # Registry + provider implementations

│   │   ├── stores/                 # Zustand transcript/settings store

│   │   ├── utils/                  # Audio, storage, provider, local-runtime helpers

│   │   └── i18n/                   # UI translations

├── local-runtimes/

│   └── whisper_cpp/                # Optional packaged whisper.cpp runtime assets

├── scripts/                        # Runtime staging/fetching and asset scripts

├── server/                         # Standalone proxy server for debugging / experiments

└── package.json

```



## 🔧 Tech Stack



| Layer | Technology |

|-------|------------|

| Desktop app | Electron 40 |

| Frontend | React 18 + TypeScript + Vite |

| Styling | Tailwind CSS |

| State | Zustand |

| Desktop services | Express + ws inside Electron |

| ASR backends | Soniox V4, Volcengine, OpenAI-compatible local ASR, `whisper.cpp` |

| Packaging | electron-builder |



## ⌨️ Keyboard Shortcut



| Shortcut | Function |

|----------|----------|

| `Ctrl+Shift+D` / `Cmd+Shift+D` | Show or hide the main window |



## 🔧 Extending Providers



1. Add a provider implementation under `frontend/src/providers/implementations/`.

2. Define accurate `ASRProviderInfo` metadata, required fields, and capability flags.

3. Register the provider in `frontend/src/providers/registry.ts`.

4. Add config-test logic in `frontend/src/utils/providerConfigTest.ts` if the provider supports validation.

5. For local-service or local-runtime flows, wire model/runtime helpers in `frontend/src/utils/localModelSetup.ts` or `frontend/src/utils/localRuntimeManager.ts`.

6. If the provider needs custom headers or native process control, extend `electron/main.ts`. Mirror that behavior in `server/` only if you still need a standalone proxy path.



## ⚠️ Notes



1. **System requirements**: Windows 10+, macOS 13+, or Linux with PulseAudio loopback support.

2. **Volcengine proxy**: normal desktop usage does not require a separate backend process; Electron starts the proxy internally.

3. **Local OpenAI-compatible mode**: discovery expects both `/v1/models` and `/v1/audio/transcriptions`.

4. **`whisper.cpp` mode**: packaged binaries are optional; users can also import or download binaries and models at runtime.

5. **Tray behavior**: closing the main window minimizes to tray; use the tray menu to exit fully.

6. **Auto-launch**: currently supported on Windows and macOS.

7. **Auto-update**: supported on Windows, macOS, and Linux AppImage builds.



### 🛡️ Windows SmartScreen Warning



Windows may show a SmartScreen warning the first time you launch DeLive. That is expected for unsigned or newly distributed apps.



1. Click **More info**.

2. Click **Run anyway**.



You can also inspect the source code directly and verify released binaries independently.



## 📄 License



Apache License 2.0



## 🙏 Acknowledgments



- [Soniox](https://soniox.com) for realtime speech recognition APIs

- [Volcengine](https://www.volcengine.com) for Chinese-focused speech recognition

- [Ollama](https://ollama.com) for local model workflows

- [`whisper.cpp`](https://github.com/ggml-org/whisper.cpp) for local open-source runtime support

- [BiBi-Keyboard](https://github.com/BryceWG/BiBi-Keyboard) for multi-provider architecture inspiration



---






[![Star History Chart](https://api.star-history.com/svg?repos=XimilalaXiang/DeLive&type=date&legend=top-left)](https://www.star-history.com/#XimilalaXiang/DeLive&type=date&legend=top-left)



**Made by [XimilalaXiang](https://github.com/XimilalaXiang)**