Ayagikei / LifeUp

Pure and highly customizable gamified to-do and habit-forming app.
196 stars 4 forks source link

支持语音输入来新增事项 #167

Open luoxiaopeng opened 1 week ago

luoxiaopeng commented 1 week ago

你的需求是否与某个问题相关联?请描述你遇到了什么问题。 支持语音输入来新增事项

描述你的解决方案 不需要支持复杂的语言功能,仅仅支持简单的,比如 晚上9点做家务3金币,5经验,如果没有识别到金币,就用设置里默认的。

描述可选方案 如果可以用一段文字来新增事项,就更好了,可以语言录入,然后手动修改一些识别错误的地方

额外信息 因为现在越来越多的App都在支持AI/语音功能,我已经很习惯用语音来创建系统的日程。相信也有很多人都在使用类似的功能。 语音新增事项,明显要快于手动点击。可以大幅提高App的使用率。当然,你可以把这个功能做成会员功能,我也是付费会员。

Issues-translate-bot commented 1 week ago

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿


Title: Supports voice input to add new items

Is your need related to a problem? Please describe what problem you encountered. Support voice input to add new items

Describe your solution There is no need to support complex language functions, only simple ones, such as 3 gold coins and 5 experience for doing housework at 9pm. If no gold coins are recognized, use the default in the settings.

Describe options none

Additional information Because more and more apps now support AI/voice functions, I have become accustomed to using voice to create system schedules. I believe many people are using similar functions. Adding new items by voice is obviously faster than manual clicking. It can greatly increase the usage of App. Of course, you can make this function a membership function. I am also a paying member.

Ayagikei commented 1 week ago

你好,感谢你的建议!

为了app的简洁性,我们不计划引入语音输入 ———— 我们建议你使用输入法的相关能力实现。

至于语言识别,在线的 AI 模型会引入较高的成本,这不太适合一次性付费模型的应用。我们曾经拉取过一个分支引入谷歌的机器学习模型来识别文本中的时间并提供选项,但之前测试下来发现效果一般,暂时还没合并到发布分支。

我们后续也会继续关注是否有离线的解决方案。

Issues-translate-bot commented 1 week ago

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿


Hi, thanks for your advice!

For the simplicity of the app, we do not plan to introduce voice input - we recommend that you use the related capabilities of the input method.

As for language recognition, online AI models will introduce higher costs, which is not suitable for the application of one-time payment models. We once pulled a branch and introduced Google's machine learning model to identify time in text and provide options. However, previous tests found that the effect was average and it has not been merged into the release branch yet.

We will continue to pay attention to whether there are offline solutions in the future.