导入资料、试听，或生成并保存

• 新建工作区，导入 PDF、EPUB、TXT、Markdown、照片或扫描页，也可以粘贴文字。
• 同一工作区可以放多份资料。原文件、整理后的文字和音频是分开的。
• 提取结果可以改。需要留一份在 App 外面时，从资料或工作区导出。

• Create a workspace and import a PDF, EPUB, text file, Markdown, photo or scan, or paste text.
• One workspace can hold several sources. Originals, edited text and audio stay separate.
• Correct the extracted text. Export a source or the workspace when you want a copy outside the app.

• 打开「工具」里的文字转语音。输入文字，选择男声或女声。
• 「试听」只播放，不写入最近转换。
• 「生成并保存」会留下可在音频库播放的文件。重启后仍然可以听。
• 中英文男声、女声已经内置，不用另装音色包。系统声音只列出这台设备实际安装的项目。
• 音频库里的播放、倍速、循环、睡眠定时和锁屏控制都只读已保存的文件，不会重新合成。

• Open Text to Audio in Tools. Enter text and choose a male or female voice.
• Listen plays the passage and does not add a recent conversion.
• Create and save keeps a file the audio library can play, including after a restart.
• Chinese and English male and female voices are bundled. System voices list only voices installed on the device.
• Resume, speed, repeat, the sleep timer and Lock Screen controls read the saved file. They do not synthesize again.

在工具里选择截图、照片或文件。真机还可以拍照和连续扫描。识别完成后对照原图修改，再复制、导出，或主动放进工作区。复杂版式和公式请以原文为准。相机功能需要有相机的设备。

In Tools, choose a screenshot, photo or file. On a device with a camera you can also capture a page or scan several pages. Compare the result with the original, then copy, export, or add it to a workspace yourself. Check complex layouts and formulas against the source. Camera actions need a device that has a camera.

工具页有离线捷径教程。在 iOS 快捷指令中查找 DocsToSpeech 的「离线识别文字」和「生成离线音频」。准备过程会打开 App，使用和工具页相同的本地能力。

The Tools tab includes an offline Shortcuts guide. In the Shortcuts app, look for DocsToSpeech actions Recognize text offline and Create offline audio. Preparation opens the app and uses the same local tools.

短句和普通句子都可以试听。如果弹出「暂时未完成」，把标题下面的原文，连同设置里的构建号、设备和操作步骤，发到 [email protected]。内存不够时，工具页会说明当前跑不起来的项目；内存够用时不显示这块说明。首次合成需要一些时间，这和安装包里的语音模型有关。

Short phrases and ordinary sentences can both be previewed. If “Could not finish” appears, send the sentence under the title, plus the build number in Settings, the device and the steps, to [email protected]. When memory is too low, Tools explains which bundled tool cannot start. That card stays hidden when the tools can run. The first synthesis takes time because the speech model is inside the app.