Ecosyste.ms: Awesome

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

Awesome Lists | Featured Topics | Projects

https://github.com/rerender2021/heard

A simple subtitle generator powered by whisper & avernakis react.
https://github.com/rerender2021/heard

avernakis desktop react subtitle whisper windows

Last synced: about 1 month ago
JSON representation

A simple subtitle generator powered by whisper & avernakis react.

Host: GitHub
URL: https://github.com/rerender2021/heard
Owner: rerender2021
License: mit
Created: 2023-03-25T08:49:54.000Z (over 1 year ago)
Default Branch: main
Last Pushed: 2023-04-22T14:53:06.000Z (over 1 year ago)
Last Synced: 2024-08-01T20:53:03.665Z (4 months ago)
Topics: avernakis, desktop, react, subtitle, whisper, windows
Language: TypeScript
Homepage: https://rerender2021.github.io/products/heard/
Size: 13.7 MB
Stars: 64
Watchers: 1
Forks: 7
Open Issues: 7
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

awesome - rerender2021/heard - A simple subtitle generator powered by whisper & avernakis react. (TypeScript)

README

[![build](https://github.com/rerender2021/heard/actions/workflows/build.yml/badge.svg?branch=main&event=push)](https://github.com/rerender2021/heard/actions/workflows/build.yml) [![pack](https://github.com/rerender2021/heard/actions/workflows/pack.yml/badge.svg?branch=main&event=push)](https://github.com/rerender2021/heard/actions/workflows/pack.yml)

# 简介

听到了! (Heard) 是一个字幕生成器，原理：

- 使用 [Whisper](https://github.com/openai/whisper) 生成字幕，支持 GPU 模式
- GUI 部分则是使用 [Ave React](https://qber-soft.github.io/Ave-React-Docs/) 开发的

![heard-usage](./docs/images/heard-usage.png)

使用它可以生成 srt 格式字幕：

![srt-demo](./docs/images/srt-demo.png)

演示视频见:

- [听到了！Whisper & GPU 离线字幕提取](https://www.bilibili.com/video/BV1FN411N7a5/)

# 使用说明

- 软件首页：https://rerender2021.github.io/products/heard/

# 开发者向

## 本地开发

```bash
> npm install
> npm run dev
```

开发过程中需要确保本机启动了 Whisper 服务器: https://github.com/rerender2021/Whisper-API/releases/tag/1.0.0

下载它，并解压到项目下，确保项目目录结构如下：

```
- whisper-gpu-server
- Whisper-API.exe
- ...
- src
- ...
- package.json
```

- 此外，还需要安装 cuda 环境，注意版本：11.7：https://developer.nvidia.com/cuda-11-7-0-download-archive?target_os=Windows&target_arch=x86_64

验证 cuda 安装成功：命令行输入 `nvcc -V`，看输出：

```bash
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_19:00:59_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0
```

- 运行 Whisper 还需要安装 ffmpeg，验证是否安装成功：`ffmpeg -v`

```bash
ffmpeg version 2022-02-10-git-b6bb6b9f22-full_build-www.gyan.dev Copyright (c) 2000-2022 the FFmpeg developers
built with gcc 11.2.0 (Rev7, Built by MSYS2 project)
...
```

## 功能扩展

默认支持 Whisper 类型为 base 的模型（英语 & 多语言），如果需要使用其它模型，可自行下载，并放在 `whisper-gpu-server\model` 下。

- 模型下载地址：https://github.com/openai/whisper/discussions/63#discussioncomment-3798552

## 打包发布

- 生成 exe

```bash
> npm run release
```

# 开源协议

[MIT](./LICENSE)

# 赞赏

`:)` 如果此软件值得赞赏，可以请作者看小说，一元足足可看八章呢。