Releases: buxuku/SmartSub
Releases · buxuku/SmartSub
2.1.0
feat
- 可自定义缓存目录
Can customize the cache directory
perf
- 优化字幕重复的问题 #158
Subtitle repetition issue #158
- 提高 ai 返回结果的兼容性 #157 #160
Improve AI return result compatibility #157 #160
- 采用标准的时间轴格式 #156
Using the standard timeline format #156
fix
- deeplx 翻译失败 #121
deeplx translation failed #121
2.0.4-beta.0
feat
- 支持调整 max_context 参数,试验性解决字幕重复的问题
Supports adjusting the max_context parameter to experimentally solve the problem of subtitle duplication
2.0.3
feat
- Ollama 服务支持直接选择本地已经安装的模型
Ollama service support allows you to directly select a locally installed model.
- 新增 AI 聚合平台 DeerAPI 支持
The new AI aggregation platform DeerAPI supports
perf
2.0.2
perf
- 优化批量任务执行逻辑
Optimize batch task execution logic
fix
- Mac 平台使用量化模型闪退
Mac platform crashes when using quantitative model
2.0.1
feat
- ollama 接口调整为使用
/chat
接口
ollama interface adjusted to use /chat
interface
fix
- windows 版本生成字幕提示词不生效
The subtitle prompts generated in Windows version are not effective
2.0.0
feat
- 字幕提取支持设置提示语,可以通过提示词优化字幕的提取结果,以及可解决繁体中文的问题
Subtitle extraction supports setting prompts, which can optimize the extraction results of subtitles through keywords, and can also solve the problem of traditional Chinese characters.
Perf
- 优化 AI 翻译,提高翻译结果条目的准确性
Optimize AI translation and improve the accuracy of translation results entries.
2.0.0-beta.6
feat
- 任务执行支持展示执行的进度
Task execution supports displaying the progress of execution.
- 增加任务执行失败的提示
Added a message indicating task execution failure
- 新增设置翻译失败重试次数
New setting: translation failed, retry times
- 可设置保存音频文件
Can be set to save audio files
perf
- 在苹果 M 芯片电脑上,非量化模型,增加对 encoder.mlmodelc 文件的检测
On Apple M-chip computer, non-quantized model, increased detection of encoder.mlmodelc file
2.0.0-beta.5
feat
- 翻译服务配置时,支持测试配置是否正确
The translation service can be configured to test whether the configuration is correct
fix
- 清除缓存失效
Clear cache invalidation
perf
- 模型下载展示更精细的进度
The model download shows a more detailed progress
2.0.0-beta.4
feat
- 提供 cuda 兼容版本
Cuda-compatible versions available
- 模型支持所有 whisper.cpp 系列模型
Models support all models in the whisper.cpp family
fix
- 下载模型进度展示问题 #120
Download model progress display problem #120
2.0.0-beta.3
feat
- 缓存音频文件,二次执行速度更快
Cache audio files for faster secondary execution
- 支持暗黑模式 #123
Dark mode is supported #123
- 支持导入 mlmodelc 文件
Import mlmodelc file is supported
fix
- 在 windows 系统下,中文路径文件无法生成字幕 #126
Under the Windows system, Chinese path files cannot generate subtitles. #126