Skip to content

Commit

Permalink
update
Browse files Browse the repository at this point in the history
  • Loading branch information
@Huanshere
Huanshere committed Sep 15, 2024
1 parent 8b48e79 commit eda2295
Show file tree
Hide file tree
Showing 2 changed files with 17 additions and 20 deletions.
19 changes: 9 additions & 10 deletions docs/install_locally_en.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,30 +6,29 @@ VideoLingo offers multiple Whisper solutions for the speech recognition text ste
|:---------|:-----------|:--------------|
| **whisper_timestamped** | • Runs locally<br>• Easy installation<br>• Uses native Whisper model | • Ideal for English only<br>• Requires GPU with 8GB+ VRAM |
| **whisperX** | • Runs locally<br>• Based on faster-whisper, excellent performance<br>• Good multi-language support | • Requires CUDA and cuDNN installation<br>• Separate wav2vec model download for each language<br>• Requires GPU with 8GB+ VRAM |
| **whisperX_api** (🌟Recommended) | • Uses Replicate API, no local computing power needed | • Replicate service may be unstable, occasional CUDA errors<br>• large-v3 model used may have inferior punctuation compared to v2 |
| **whisperX_api** (🌟Recommended) | • Uses Replicate API, no local computing power needed | • The large-v3 model used may have inferior punctuation compared to v2 |

## 📋 Preparation

1. Obtain the `API_KEY` for `claude-3-5-sonnet`. Recommended affordable channel: [Yunwu API](https://api2.wlai.vip/register?aff=TXMB), only ¥35/1M, 1/3 of the official price. Of course, you can also use other API providers, but it's recommended to choose `claude-3-5-sonnet` > `Qwen 1.5 72B Chat` > `deepseek-coder`
1. Obtain the `API_KEY` for `claude-3-5-sonnet`. Recommended affordable channel: [Yunwu API](https://api2.wlai.vip/register?aff=TXMB), only ¥15/1M, 1/8 of the official price. Of course, you can also use other API providers, but it's recommended to choose `claude-3-5-sonnet` > `Qwen 1.5 72B Chat` > `deepseek-coder`

![yunwu](https://github.com/user-attachments/assets/7aabfa87-06b5-4004-8d9e-fa4a0743a912)

2. If using `whisperX_api`, please register and set up payment method on the [Replicate official website](https://replicate.com/account/api-tokens) to obtain your token. You can also contact me in the QQ group for free test tokens.
## 💾 One-Click Package Download

If you don't want to install manually, we also provide a `whisperX_api` version of the Windows one-click integrated package:
## 💾 One-Click Package

1. Download the `v0.7.0` one-click integrated package (700M): [Direct Link](https://vip.123pan.cn/1817874751/8026614) | [Baidu Pan Backup](https://pan.baidu.com/s/1H_3PthZ3R3NsjS0vrymimg?pwd=ra64)
We provide a Windows one-click integrated package using `whisperX_api`:

2. Extract the downloaded compressed file to your desired location
1. Download the `v0.7.0` one-click package (700M): [Direct Link](https://vip.123pan.cn/1817874751/8026614) | [Baidu Pan Backup](https://pan.baidu.com/s/1H_3PthZ3R3NsjS0vrymimg?pwd=ra64)

3. Double-click `一键启动.bat` in the extracted folder
2. After extracting, double-click `OneKey.bat` in the folder

4. In the opened browser window, follow the interface prompts for configuration and usage
3. In the opened browser window, make necessary configurations in the sidebar, then create your video with one click!

> Note: Refer to the image at the bottom for the key configuration process in the web interface
> Note: Refer to the image at the bottom for the key configuration in the sidebar
## 🛠️ Manual Installation Process (Windows)
## 🛠️ Source Code Installation Process (Windows)

### Prerequisites

Expand Down
18 changes: 8 additions & 10 deletions docs/install_locally_zh.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,31 +6,29 @@ VideoLingo 语音识别文本步骤提供多种 Whisper 方案的选择（因为
|:-----|:-----|:-----|
| **whisper_timestamped** | • 本地运行<br>• 安装简便<br>• 使用原生 Whisper 模型 | • 仅英文效果理想<br>• 需要8G以上显存的显卡 |
| **whisperX** | • 本地运行<br>• 基于 faster-whisper，性能卓越<br>• 多语言支持好 | • 需安装 CUDA 和 cuDNN<br>• 各语言需单独下载 wav2vec 模型<br>• 需要8G以上显存的显卡 |
| **whisperX_api** (🌟推荐) | • 利用 Replicate API，无需本地算力 | • Replicate 服务可能不稳定 偶发 CUDA 错误<br>• 使用的large-v3 标点效果可能不如v2 |
| **whisperX_api** (🌟推荐) | • 利用 Replicate API，无需本地算力 | • 使用的large-v3 标点效果可能不如v2 |

## 📋 前期准备

1. 获取 `claude-3-5-sonnet` 的 `API_KEY`，推荐便宜渠道：[云雾API](https://api2.wlai.vip/register?aff=TXMB)，仅仅 ￥35/1M，官方价格的 1/3。当然这一步你也可以换成别的api提供商，但仅仅建议选用 `claude-3-5-sonnet` > `Qwen 1.5 72B Chat` > `deepseek-coder`
1. 获取 `claude-3-5-sonnet` 的 `API_KEY`，推荐便宜渠道：[云雾API](https://api2.wlai.vip/register?aff=TXMB)，仅仅 ￥ 15/1M，官方价格的 1/8。当然这一步你也可以换成别的api提供商，但仅仅建议选用 `claude-3-5-sonnet` > `Qwen 1.5 72B Chat` > `deepseek-coder`

![yunwu](https://github.com/user-attachments/assets/7aabfa87-06b5-4004-8d9e-fa4a0743a912)

2. 若选用 `whisperX_api`，请在 [Replicate官网](https://replicate.com/account/api-tokens) 注册并绑定支付方式，获取你的令牌。也可在 QQ 群联系我免费提供测试用。

## 💾 一键包下载
## 💾 一键整合包

如果你不想手动安装,我们也提供了 `whisperX_api` 版本的 Windows 一键整合包:
我们提供了使用 `whisperX_api` 的 Windows 一键整合包:

1. 下载 `v0.7.0` 一键整合包(700M): [直达链接](https://vip.123pan.cn/1817874751/8026614) | [度盘备用](https://pan.baidu.com/s/1H_3PthZ3R3NsjS0vrymimg?pwd=ra64)

2. 解压下载的压缩文件到想要的位置
2. 解压后双击运行文件夹中的 `一键启动.bat`

3. 双击运行解压后文件夹中的 `一键启动.bat`
3. 在打开的浏览器窗口中，在侧边栏进行必要配置，然后一键出片！

4. 在打开的浏览器窗口中,按照界面提示进行配置和使用
> 提示: 侧边栏配置 key 的说明可以参考最下方图片
> 提示: 网页中配置 key 的过程参考最下方图片
## 🛠️ 手动安装流程 (Windows)
## 🛠️ 源码安装流程 (Windows)

### 前置依赖

Expand Down

0 comments on commit eda2295

Please sign in to comment.