Москалькова заявила о новых условиях Киева для возвращения россиян с территории Украины14:51
Finally, a note around TTS. On Android TTS is not build into AOSP. [6] Instead you need to have an application that provides TextToSpeechService. On most devices this is usually handled by Google's Speech Recognition & Synthesis app, but that is of course not installed by default on GrapheneOS. There are FOSS apps like Sherpa TTS, but the models they provide only support one language at a time and the app can only load one model at a time. This limitation means Sherpa cannot easily switch between different languages, which is exactly what you need for a translation app. I think espeak NG might work, but I haven't given it a try considering the voice quality.
，更多细节参见快连下载安装
这方面，国内的光帆科技（Lightwear）其实已经给出了先行版本——他们在耳机上装了一对摄像头，连续不间断地捕捉现实世界，但这些信息在 AI 使用之后便「阅后即焚」，既保证可用性，又具备私密性——这套方案，是不是也有点果味儿？
当然，影像也是Robot Phone上的一大看点。除了防抖和两亿像素之外，在发布会上，荣耀还宣布与影像设备品牌阿莱（ARRI）合作，双方将在色彩科学、电影视觉、专业制作流程等方面展开合作。未来，这一系列成果将在Robot Phone上落地。
That’s the theory, anyway. But current LLM APIs are expensive, and I wasn’t about to spend thousands on a weekend project. So I got creative—and abused leveraged multiple low-cost or free API channels: