XRTranslate v0.3.5

下载 / Download

Windows · ZIP	Linux · ZIP
下载 Windows x64	下载 Linux x64

解压后运行包内的 XRTranslate-v0.3.5 程序。
Extract the ZIP and launch XRTranslate-v0.3.5.

---

以下为相较于 v0.3.4 正式版 的主要变化。
Changes since the v0.3.4 stable release.

新增 / New

• 账号登录：可登录豆包使用语音识别，登录 ChatGPT 使用翻译及所选模型支持的对话功能；支持添加、切换和退出多个账号。/ Account sign-in: use Doubao for speech recognition and ChatGPT for translation and supported chat features, with controls to add, switch, and sign out of accounts.
• 在线音频与播客：影音支持公开播客、RSS 节目、Apple Podcasts 和小宇宙链接；会议支持网络电台、直播音频及 RSS 中的直播节目。加载链接后即可选择节目。/ Online audio and podcasts: import public podcasts, RSS episodes, Apple Podcasts, and Xiaoyuzhou links in Media. Meetings support internet radio, live audio, and live items in RSS feeds. Load a link and choose a program.
• 可爱翻译风格：Prompt Studio 新增“可爱”内置卡片，让译文表达更轻快、柔和，可用于不同目标语言。/ Cute translation style: a new built-in Prompt Studio card gives translations a warmer, more playful tone across target languages.
• 英语识别模型：新增 Bark 3.0 本地英语语音识别模型选项。/ English speech recognition: Bark 3.0 is now available as a local English recognition option.

变化 / Changes

• 播放器：音轨切换整合进播放控制栏；音频处理使用选中的音轨。全屏展开菜单更通透，字幕、音量和音轨操作更集中。/ Player controls: switch audio tracks directly from the playback bar and process the selected track. Fullscreen menus are more transparent, with subtitle, volume, and track controls kept together.
• 字幕呈现与导出：可调整字幕字体、字号和颜色，Windows、Linux 支持导出带字幕的 MP4 视频。/ Subtitle appearance and export: adjust subtitle fonts, sizes, and colors, and export subtitled MP4 videos on Windows and Linux.
• 会议与连续翻译：改进会议音源设置、录音回听和音量控制，改善连续说话时的句子衔接与翻译等待。/ Meetings and continuous translation: improved audio-source setup, recording playback, and volume controls, with better sentence continuity and less waiting during continuous speech.
• 模型配置：欢迎页按本地模型、账号登录、在线 API 排列；账号选择与下载进度显示更简洁。/ Model setup: onboarding now presents local models, account sign-in, and online APIs in that order, with simpler account selection and download progress.

修复 / Fixes

• 修复媒体播放时滚动侧栏或切换页面造成的明显卡顿，以及部分视频字幕未显示的问题。/ Fixed severe slowdowns when scrolling the sidebar or changing pages during media playback, and missing subtitles in some videos.
• 修复部分语音服务在静音或无识别结果时中断任务的问题，改善长时间流式识别的稳定性。/ Fixed tasks stopping when some speech services returned no text or encountered silence, and improved long-running streaming recognition.
• 修复纯在线服务配置仍要求本地模型的问题，改善阿里语音识别与翻译服务的使用体验。/ Fixed online-only setups incorrectly requiring local models, and improved Alibaba speech recognition and translation.
• 改善阿里翻译服务对上下文和 Prompt Studio 翻译风格的使用，修复部分识别收尾和字幕衔接问题。/ Improved how Alibaba translation uses context and Prompt Studio styles, and fixed issues with recognition completion and subtitle continuity.

使用提示 / Getting started

首次启动时，跟随欢迎页面选择本地模型、登录账号或在线服务。
Follow the welcome screen to choose local models, sign in to an account, or configure an online service.

Windows 可在影音页安装播放组件；Linux（Debian/Ubuntu）可用 sudo apt install libmpv2 libwebkit2gtk-4.1-0 安装视频播放与豆包登录所需组件。在线节目能否播放取决于来源是否公开可访问及其链接是否有效。
Install Windows playback support from Media. On Debian/Ubuntu, use sudo apt install libmpv2 libwebkit2gtk-4.1-0 for video playback and Doubao sign-in. Online programs require a publicly accessible source and a valid link.

游戏组件与安装说明见 XRT-Avatar-Plugin。
Find game components and setup instructions at XRT-Avatar-Plugin.

XRTranslate v0.3.4

下载 / Download

Windows · ZIP	Android · APK	Linux · ZIP
下载 Windows x64	下载 Android ARM64	下载 Linux x64

Windows / Linux：解压后运行包内的 XRTranslate-v0.3.4 程序；Android：下载并安装 APK。
Windows / Linux: extract the ZIP and launch XRTranslate-v0.3.4; Android: download and install the APK.

---

以下为相较于 v0.3.3 正式版 的主要变化。
Changes since the v0.3.3 stable release.

新增 / New

• 游戏插件（已测试VRChat）：集中配置 VRChat、ChilloutVR、Overte、Resonite、Second Life 与 Big Walk。配套组件提供角色字幕及游戏内翻译控制；字幕面板可独立开关，支持自然大字排版与静态页头页脚；Big Walk 可直接发送语音译文和文字到聊天。/ Game Plugins(Tested in VRC): configure VRChat, ChilloutVR, Overte, Resonite, Second Life, and Big Walk in one place. Companion components offer avatar captions and in-game translation controls where supported, with a separate panel toggle, natural large-text layout, and static headers and footers; Big Walk supports translated speech and text in chat.
• ChilloutVR OSC：新增原生聊天气泡输出，可与 VRChat 分别保存设置。/ ChilloutVR OSC: send translations to native chat bubbles, with settings saved separately from VRChat.
• 字幕编辑：媒体字幕可修改原文、译文和时间、单句重译或删除；媒体与会议字幕可将相邻条目拖动合并，修改自动保存。/ Subtitle editing: edit media subtitle text, translations, and timing, retranslate individual cues, or delete them. Drag neighboring media or meeting entries together to merge; changes save automatically.
• 会议回听：播放录音、拖动进度、从指定发言开始回听，并切换不同来源或续录的录音。/ Meeting playback: play and seek recordings, listen from a selected utterance, and switch between recordings from different sources or continuations.
• 就地查词：翻译历史和大字演示中选中文字即可查词、复制或收藏，也可切换查询语言。/ Inline lookup: select text in translation history or presentation mode to look it up, copy it, or save it, with a choice of target language.
• SteamVR 语音交互：手腕上的编辑、发送按钮配合 3D 空间形象，可审核待发送内容并用语音修改。/ SteamVR voice interaction: wrist-mounted Edit and Send controls work with the 3D companion to review messages and edit them by voice.
• Linux 虚拟麦克风：自动创建 XRTranslate Microphone，可混合麦克风、应用声音和语音播报，供其他应用使用。/ Linux virtual microphone: automatically create XRTranslate Microphone to mix microphone input, application audio, and synthesized speech for other apps.

变化 / Changes

• 影音播放：Windows、Linux 支持高性能页内视频播放，使用统一的图标式播放、音量、字幕及全屏控制；全屏按钮悬浮在画面上并自动隐藏。字幕时间轴可直接定位，手机排版与滑动浏览更舒适。/ Media playback: high-performance video playback inside the app on Windows and Linux, with compact playback, volume, subtitle, and fullscreen icons. Fullscreen controls float over the video and hide automatically. Seek from the subtitle timeline, with improved mobile layout and scrolling.
• 翻译与字幕：更好地利用上下文衔接句意、处理完整句子和修正识别错误；继续会议和单句重译也能参考已有内容。自动识别模式会显示实际选用的模式。/ Translation and subtitles: better use of context for coherent meaning, complete sentences, and recognition corrections, including resumed meetings and individual-cue retranslation. Automatic recognition mode shows the mode actually selected.
• SteamVR 字幕：默认单条居中显示，原文与译文一起出现，紧邻消息可合并呈现；多条模式支持平滑滚动。预览内可直接调整背景、透明度和边框颜色，并可隐藏说话人编号。/ SteamVR captions: one centered caption by default, with source and translation shown together and closely spaced messages combined. Multiple captions scroll smoothly. Adjust background, transparency, and border colors in the preview, and optionally hide speaker numbers.
• OSC 与歌词：歌名、歌手和同步歌词可分开显示，支持歌词时间微调；VRChat 麦克风开关可与识别联动。/ OSC and lyrics: display track metadata separately from synchronized lyrics and adjust lyric timing. VRChat microphone controls can also pause or resume microphone recognition.
• 单词本：支持搜索词语及译文、另存到其他单词本、分别管理不同译法，并可重新生成例句和查看例句翻译。/ Vocabulary books: search words and translations, save entries to other books, manage alternative translations, and regenerate or translate examples.
• 使用体验：欢迎页增加游戏功能轮播，模型与服务配置、连接提示和功能说明更简洁；会议记录的续录、浏览、重点定位及导出更方便。/ Everyday use: onboarding adds a game-feature showcase, with clearer model and service setup, connection feedback, and shorter guidance. Meeting continuation, browsing, highlight navigation, and export are easier to use.

修复 / Fixes

• 改善连续语音和流式识别的漏字、重复、断句与收尾处理，减少不必要的等待，提升字幕连续性。/ Improved missing or repeated text, sentence boundaries, and ending detection in continuous and streaming speech, reducing unnecessary waits and improving subtitle continuity.
• 改善 Linux 音频播放、Android 语音功能及部分在线服务的稳定性；本地模型启动失败时提供更明确的提示。/ Improved Linux audio playback, Android speech features, and reliability with some online services, with clearer messages when local models fail to start.

使用提示 / Getting started

首次启动时，跟随欢迎页面选择模型或在线服务。Windows 可在影音页安装播放组件；Linux 视频播放需系统 MPV 运行库（Debian/Ubuntu：libmpv2），虚拟麦克风需 PulseAudio 或 PipeWire 的 PulseAudio 兼容服务。
Follow the welcome screen to choose models or online services. Install Windows playback support from Media. Linux video playback requires the MPV runtime (libmpv2 on Debian/Ubuntu); the virtual microphone requires PulseAudio or PipeWire's PulseAudio compatibility service.

游戏组件与安装说明见 XRT-Avatar-Plugin。SteamVR 空间语音交互及发送前审核默认关闭；语音修改需使用支持对话的模型。
Find game components and setup instructions at XRT-Avatar-Plugin. SteamVR spatial voice interaction and message review start disabled; spoken editing requires a chat-capable model.

XRTranslate v0.3.4-beta.1

下载 / Download

Windows · ZIP	Linux · ZIP
下载 Windows x64	下载 Linux x64

解压后运行包内的 XRTranslate-v0.3.4-beta.1 程序。本次为测试版，仅提供 Windows 和 Linux 桌面包，不提供 Android APK。
Extract the ZIP and launch XRTranslate-v0.3.4-beta.1. This beta provides Windows and Linux desktop packages only; no Android APK is included.

---

新增 / New

• 新增“游戏插件”页面，集中配置 VRChat、ChilloutVR、Overte、Resonite、Second Life 与 Big Walk 的字幕和翻译功能。角色字幕可显示原文、译文及歌曲信息，部分游戏支持在游戏内控制翻译。这些功能仍处于测试阶段，需安装对应配套组件。/ Added a Game Plugins page for subtitle and translation features in VRChat, ChilloutVR, Overte, Resonite, Second Life, and Big Walk. Avatar captions can show original text, translations, and song information, with in-game translation controls in some games. These features are experimental and require their companion components.
• 新增 ChilloutVR OSC 聊天气泡输出，设置方式与 VRChat OSC 一致，两款游戏可分别保存配置。/ Added ChilloutVR OSC chat bubbles, with familiar VRChat OSC controls and separate settings for each game.
• 支持向 Big Walk 游戏聊天发送语音译文、打字译文或直接输入的文字，需安装配套游戏组件。/ Send speech translations, translated typed messages, or direct text to Big Walk chat using its companion component.
• Windows SteamVR 新增可抓取的空间麦克风，靠近嘴部即可开启识别；支持发送前确认，以及用语音修改待发送的消息。/ Added a grabbable spatial microphone in SteamVR on Windows: bring it near your mouth to start recognition. Review messages before sending and edit them by voice.
• Linux 可自动创建虚拟麦克风，在音频工作室混合麦克风、应用声音和语音播报，再传给其他应用。接收端选择 XRTranslate Microphone 即可使用。/ Linux can now create a virtual microphone automatically. Mix your microphone, application audio, and synthesized speech in Audio Studio, then select XRTranslate Microphone in the receiving app.
• 媒体字幕支持编辑原文、译文和起止时间，以及单句重新翻译。双击文字或时间即可编辑，也可从某条字幕直接开始播放。/ Edit subtitle text, translations, and timestamps, or retranslate a single cue. Double-click text or a timestamp to edit, or start playback from a subtitle.
• 媒体和会议字幕可拖到相邻字幕上快速合并，松手后确认，保留目标字幕的说话人等信息并立即保存。/ Drag a media or meeting subtitle onto its neighbor to merge after confirmation, keeping the destination's speaker and other details and saving immediately.
• 会议记录新增录音播放和进度拖动，可从某条发言开始回听，也可切换不同来源或续录的录音。/ Play and seek through meeting recordings, listen from a selected utterance, and switch between recordings from different sources or continuations.
• 翻译历史和演示模式支持选中文字后就地查词、复制和收藏，查询时可切换目标语言。/ Select text in translation history or presentation mode to look it up, copy it, or save it to vocabulary, with a choice of target language.

变化 / Changes

• 媒体翻译和单句重译可参考前后文；继续会议时，也会衔接已有发言，帮助理解上下文。/ Media translation and single-cue retranslation can use surrounding context. Resumed meetings also refer to earlier utterances.
• 大量字幕浏览更流畅，识别过程中列表更稳定，跟随播放时滚动更平滑。媒体字幕导航条会突出播放位置附近的内容，查找和跳转更方便。/ Long subtitle lists are more responsive and stay steadier during recognition, with smoother playback following. The media navigation bar emphasizes nearby subtitles for easier browsing and seeking.
• OSC 页眉和页脚可同时选择歌词：上方显示歌名与歌手，下方显示歌词，统一使用 🎵 前缀。歌词可调早或调晚，无歌词时仍可查看曲目信息。/ OSC headers and footers can both use Lyrics: show the title and artist above and lyrics below, with a 🎵 prefix. Adjust lyrics earlier or later, and keep track information available when lyrics are missing.
• VRChat 可同步麦克风开关：在 XRTranslate 开关麦可联动游戏，游戏内静音也会暂停麦克风识别，系统音频不受影响。/ Synchronize microphone controls with VRChat: toggling the microphone in XRTranslate can control the game, and muting in-game pauses microphone recognition while system audio continues.
• 游戏插件配对更方便，可一键复制连接信息；角色字幕自动适应显示空间，内容更新更连贯。/ Game pairing is easier with one-click copying of connection information. Avatar captions fit the available space and update more smoothly.
• 单词本支持搜索词语和译文、另存到其他单词本、分别管理不同译法，并可重新生成例句或查看例句翻译。/ Search words and translations in vocabulary books, save them to another book, manage translations individually, and regenerate examples or reveal their translations.
• 会议重点和备注更容易查找、定位和导出。结束录音会处理剩余内容，切换会议或导出时自动保存编辑。/ Meeting highlights and notes are easier to find, revisit, and export. Ending a recording finishes pending content, and switching meetings or exporting saves your edits.
• 欢迎页增加功能预览，语言、模型和可选功能设置更清楚；精简界面说明，保留必要操作指引。提示词工作室也更方便查看实际使用的翻译提示词。/ The welcome page adds feature previews and clearer language, model, and optional-feature setup. Interface descriptions are simpler while practical guidance remains. Prompt Studio also makes it easier to inspect the translation prompts in use.

修复 / Fixes

• 修复连续识别时可能漏字或错误拼接文字的问题，减少说完一句话后的不必要等待，并改善录音结尾的识别。/ Fixed missing or incorrectly joined text during continuous recognition, reduced unnecessary waits after complete sentences, and improved recognition at the end of recordings.
• 减少长句被不合理拆分的问题，修复不同翻译任务混用参考内容及字幕语言延续不正确的情况。/ Reduced unnecessary sentence splitting and fixed unrelated translation tasks sharing context or subtitles retaining the wrong language.
• 修复 OSC“仅译文”模式提前显示原文、挤掉已有译文的问题，改善聊天框更新体验。/ Fixed source text appearing in OSC target-only mode and displacing existing translations, and improved chatbox updates.
• 改善 Linux 音频播放稳定性与播放、暂停的响应。应用声音更容易分辨，未设置语音播报时也可正常混合其他声音。/ Improved Linux audio playback stability and play/pause responsiveness. Application audio is easier to identify, and other sounds can be mixed without setting up speech synthesis.
• 修复会议重点与发言关联不稳定的问题，并改善部分在线翻译服务的使用稳定性。/ Fixed unstable links between meeting highlights and utterances, and improved reliability with some online translation services.

使用提示 / Getting started

Linux 包适用于 Ubuntu 26.04 等使用 glibc 2.43 或更新版本的发行版；较旧系统需从源码构建。虚拟麦克风需要 PulseAudio，或启用 PulseAudio 兼容服务的 PipeWire。
The Linux package requires glibc 2.43 or newer, as in Ubuntu 26.04. Older distributions require a source build. The virtual microphone needs PulseAudio or PipeWire with PulseAudio compatibility enabled.

首次启动时，跟随欢迎页面选择模型或在线服务；可在设置中开启“接收测试版更新”。媒体字幕编辑和合并需暂停识别任务，会议合并需先结束会议。
On first launch, follow the welcome screen to choose models or online services. Enable “Receive beta updates” in Settings for beta releases. Pause media recognition before editing or merging subtitles, and end a meeting before merging its subtitles.

游戏插件的安装入口与操作指引在对应页面，也可前往 XRT-Avatar-Plugin 获取配套组件。VRChat 麦克风同步需开启游戏的 Toggle Voice；SteamVR 语音修改需使用支持对话的模型。
Find installation links and setup guidance on each Game Plugin page, or get companion components from XRT-Avatar-Plugin. Enable Toggle Voice in VRChat for microphone synchronization. SteamVR spoken editing requires a chat-capable model.

XRTranslate v0.3.3

下载 / Download

Windows · ZIP	Android · APK	Linux · ZIP
下载 Windows x64	下载 Android ARM64	下载 Linux x64

Windows / Linux：解压后运行包内的 XRTranslate-v0.3.3 程序；Android：下载并安装 APK。
Windows / Linux: extract the ZIP and launch XRTranslate-v0.3.3; Android: download and install the APK.

---

新增 / New

• 新增单词本：可从快捷翻译收藏词语，按单词本整理和编辑译法，并生成例句与回译。/ Added vocabulary books: save words from quick translation, organize and edit saved translations, and generate example sentences with back-translations.
• 新增译文演示模式，以大字展示实时译文，支持黑白背景切换和跟随最新内容。/ Added presentation mode for live translations, with larger text, black or white backgrounds, and a follow-latest option.
• VRChat OSC 聊天框的页眉或页脚现在可以显示当前歌曲的同步歌词。/ VRChat OSC chatbox headers and footers can now show synchronized lyrics for the current song.

变化 / Changes

• 安卓通知栏可快捷翻译剪贴板并自动写回结果，当前词语与收藏按钮一起显示，收藏更直观。/ Translate the clipboard from Android notifications and copy the result back automatically, with the current word shown beside its save button.
• 优化翻译机制，提升翻译效果。/ Improved translation processing and translation quality.
• 快捷翻译浮窗可直接切换目标语言，收藏与复制操作更方便。/ Switch the target language directly in the quick-translation window, with easier saving and copying.
• 影音页面统一文件导入入口，支持将文件直接拖到首页；全屏播放新增空格暂停和方向键快进、快退，控制栏可自动隐藏。/ Media now has a unified file import entry and accepts files dropped onto its home page. Full-screen playback adds Space to pause, arrow-key seeking, and auto-hiding controls.
• 会议重点更容易标记、备注和查找，导出的会议记录新增重点时间索引。/ Meeting highlights are easier to mark, annotate, and find, with a timestamped highlight index in exported notes.
• 扩充在线识别、翻译和朗读服务选择，包括 Google、Groq、硅基流动与 OpenRouter 等选项；使用时需配置自己的服务密钥。/ Expanded online recognition, translation, and speech options, including Google, Groq, SiliconFlow, and OpenRouter. These services require your own API keys.
• 术语库支持批量粘贴词条和共享领域规则，添加、纠正常见误识别与整理词汇更方便。/ The terminology library supports bulk paste and shared domain rules, making words easier to add, organize, and correct when misrecognized.

修复 / Fixes

• 修复安卓输入法在点击和拖动文本光标时反复闪烁的问题。/ Fixed the Android keyboard flickering when tapping or dragging the text cursor.
• 修复欢迎页面选择模型选项后下拉框未关闭的问题。/ Fixed model dropdowns staying open after a selection on the welcome screen.
• 修复快捷翻译浮窗的显示问题，并改善识别与译文在生成过程中的实时显示。/ Fixed quick-translation window rendering and improved live display while recognition and translation results are being generated.

使用提示 / Getting started

首次启动时，跟随欢迎页面选择模型或在线服务。音视频的可选播放组件可在影音页直接安装。
On first launch, follow the welcome screen to choose models or online services. Optional playback support can be installed from the Media page.

XRTranslate v0.3.2

新增 / New

• 新增桌面快捷翻译：在其他应用中选中或复制文字，按 Ctrl+Alt+Y 即可在小浮窗查看和复制译文；快捷键可在设置中修改或关闭。/ Added desktop quick translation: select or copy text in another app and press Ctrl+Alt+Y to view and copy the translation in a floating window. The shortcut can be changed or disabled in Settings.
• 新增翻译历史搜索，支持可选的本地保存、关键词查找、单条删除和按日期范围批量删除；本地保存默认关闭。/ Added translation history search with optional local saving, keyword search, individual deletion, and deletion by date range. Local saving is off by default.
• 媒体页面新增 SRT/VTT 字幕文件翻译，保留原时间轴，支持暂停续译及 SRT/VTT/LRC 导出。/ Added SRT/VTT subtitle-file translation in Media, preserving cue times with pause/resume and SRT/VTT/LRC export.
• VRChat OSC 与 SteamVR 字幕支持独立筛选音频来源，可选择显示麦克风、系统音频或两者的翻译结果。/ VRChat OSC and SteamVR subtitles now have independent audio-source filters for microphone, system audio, or both.

变化 / Changes

• 媒体页面统一管理音频、视频与字幕任务，媒体和会议插件默认开启。/ Media now brings audio, video, and subtitle tasks together. Media and Meeting plugins are enabled by default.
• 改善媒体播放与字幕浏览，支持拖动进度和点击字幕跳转播放；未启用播放时也可管理翻译任务和导出字幕。/ Improved media playback and subtitle browsing with seeking and click-to-play subtitles. Translation tasks and subtitle export remain available without playback.
• 优化常用界面与操作提示，轻微问题以简短通知显示，减少使用中断。/ Refined common controls and feedback, with brief notices for minor issues to reduce interruptions.

修复 / Fixes

• 修复了一些 bug。/ Fixed some bugs.

下载与运行 / Download and Launch

• Windows：下载 XRTranslate-v0.3.2-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.3.2.exe。/ Download XRTranslate-v0.3.2-win-x64.zip, extract it, and run XRTranslate-v0.3.2.exe.

XRTranslate v0.3.1

新增 / New

• 新增固定第三语种，可在双向互译或单向翻译时同时获得另一种语言的译文，且不随主翻译方向切换。/ Added a fixed third language for bidirectional and one-way translation, providing an additional translation that stays in the selected language when the main direction changes.
• 新增仅识别模式，无需翻译即可将语音转为文字，并用于字幕和文字输出。/ Added recognition-only mode to turn speech into text for subtitles and text output without translation.
• 支持通过麦克风语音指令打开或关闭 SteamVR 字幕。/ Added microphone voice commands to show or hide SteamVR subtitles.

变化 / Changes

• VRChat OSC 和 SteamVR 字幕支持同时展示第三语种译文，译文分行显示。/ VRChat OSC and SteamVR subtitles can now display the additional translation on a separate line.
• VRChat OSC 的文字翻译与翻译页面共用语种选择，切换更直观。/ Text translation in VRChat OSC now shares language selections with the Translation page for easier switching.
• 改善 SteamVR 字幕显示的流畅度与稳定性。/ Improved the smoothness and stability of SteamVR subtitles.

修复 / Fixes

• 修复了一些 bug。/ Fixed some bugs.

当前支持 / Supported Features

• SteamVR 游戏内悬浮字幕与视角跟随 / SteamVR in-game subtitles with gaze tracking
• 麦克风、系统音频与指定应用声音输入 / Microphone, system audio, and per-application audio input
• 可视化 Audio Studio 与自定义音频方案 / Visual Audio Studio and custom audio graphs
• 自动双向语音识别与翻译 / Automatic bidirectional speech recognition and translation
• 本地模型与在线服务，多种模型尺寸、显存和语种提示 / Local models and online services, with model size, VRAM, and language guidance
• NVIDIA 与 AMD 显卡上的本地识别和翻译 / Local recognition and translation on NVIDIA and AMD graphics cards
• 翻译风格卡片与自然中文风格 / Translation style cards and a natural Chinese style
• TTS 中心、内置与自定义音色、声音克隆 / TTS Center, built-in and custom voices, and voice cloning
• 对话上下文与专业术语辅助（取决于所选模型） / Conversation context and terminology assistance, depending on the selected model
• 识别与翻译历史、桌面悬浮字幕 / Recognition and translation history, and floating desktop subtitles
• VRChat OSC Chatbox，直接发送与翻译后发送 / VRChat OSC Chatbox with direct and translated text input
• 自定义 Chatbox 格式、页眉、页脚和消息时长 / Custom Chatbox formatting, headers, footers, and message duration
• VRCX 人名识别辅助与词汇图谱领域管理 / VRCX-assisted name recognition and Vocabulary Graph domain management
• 说话人识别与稳定编号 / Speaker recognition with consistent numbering
• 音视频翻译、字幕预览与导出 / Audio and video translation, subtitle previews, and subtitle export
• 会议录音、转写与记录管理 / Meeting recording, transcription, and record management
• 自动更新与用户设置保留 / Automatic updates with user-setting preservation

下载与运行 / Download and Launch

• Windows：下载 XRTranslate-v0.3.1-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.3.1.exe。/ Download XRTranslate-v0.3.1-win-x64.zip, extract it, and run XRTranslate-v0.3.1.exe.

XRTranslate v0.3.0

新增 / New

• 新增屏幕文字翻译，可在悬浮窗中选择、移动和调整识别区域，并查看原文与译文。/ Added screen text translation with a movable, resizable capture region and floating original-text and translation results.
• 翻译主界面支持直接输入文字并翻译，完成后可复制译文，无需开启麦克风。/ Added text translation directly on the Translation page, with easy copying of results and no microphone required.

变化 / Changes

• 精简翻译页面，可分别控制翻译、麦克风和系统音频输入，更方便切换日常使用方式。/ Simplified the Translation page with separate controls for translation, microphone input, and system audio input.
• 优化桌面悬浮窗，可直接控制翻译和音频输入，控制条可收起以减少遮挡，新消息展示更清晰。/ Improved the floating desktop window with direct translation and audio-input controls, a collapsible toolbar, and clearer presentation of new messages.
• 页面切换更流畅，优化窄窗口布局、常用按钮和设置操作。/ Added smoother page transitions and refined compact layouts, common controls, and settings.
• SteamVR 字幕预览支持直接拖动和缩放，更方便调整显示位置与大小。/ SteamVR subtitle previews can be dragged and resized directly for easier positioning and sizing.
• 改善实时字幕、语音播报、会议转写与记录浏览体验。/ Improved live subtitles, speech playback, meeting transcription, and record browsing.
• 优化首次设置、资源下载和软件更新流程。/ Refined first-time setup, resource downloads, and app updates.

修复 / Fixes

• 修复了一些 bug。/ Fixed some bugs.

下载与运行 / Download and Launch

• Windows：下载 XRTranslate-v0.3.0-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.3.0.exe。/ Download XRTranslate-v0.3.0-win-x64.zip, extract it, and run XRTranslate-v0.3.0.exe.

XRTranslate v0.2.11

TTS

文本转语音仍为试验功能，音质与克隆效果因音源、模型和语种而异 / TTS remains experimental; voice quality and cloning results vary by reference recording, model, and language.

新增 / New

• 新增 TTS 中心，内置 Miku 与“可爱”音色，可试听参考音频并一键使用；支持添加、搜索和管理自己的音色卡片。/ Added TTS Center with Miku and Cute voice cards, reference previews, one-click voice selection, and support for adding, finding, and managing your own voices.
• 新增音色默认自动识别参考音频文本，可选择语种，也可手动填写；选择语音模型后即可看到 TTS 中心入口。/ New voice cards transcribe reference audio automatically, with language selection and an optional manual transcript. TTS Center appears when a speech model is configured.
• 翻译风格改为简洁的卡片选择器，支持滚动查看完整风格内容，并内置适用于多种源语种的自然中文风格。/ Translation styles now use a simple card picker with scrollable descriptions and a built-in natural Chinese style for multiple source languages.
• 新增 XRTranslate 专属互动角色，支持拖动、欢迎问候和对话气泡；会根据页面与操作选择合适的位置，重新进入欢迎页时再次打招呼。/ Added an interactive XRTranslate companion with dragging, welcome greetings, and speech bubbles. It adapts its position to the page and your actions, and greets you again when you return to the welcome page.
• 支持在兼容的 AMD 显卡上使用本地语音识别与翻译，并自动匹配所需资源。/ Added local speech recognition and translation on compatible AMD graphics cards, with automatic resource selection.
• 新增批量功能检查与异常报告，方便检查不同模型和语种组合并定位问题。/ Added batch functionality checks and failure reports to help verify model and language combinations and diagnose issues.

变化 / Changes

• Audio Studio 仅保留一个完整的内置默认方案，仍可新建自己的方案；优化常用操作、音频重置和连接管理。/ Audio Studio now provides one complete built-in default graph while keeping custom graph creation, with improved common controls, audio reset, and connection management.
• 精简翻译页面与工作室操作，识别设置可按需展开，音频输入随窗口宽度自动排列；TTS 输出统一在 Audio Studio 中管理。/ Simplified translation and studio controls, with expandable recognition settings and audio inputs that adapt to the window width. TTS output is managed in Audio Studio.
• 优化 SteamVR 字幕调节、双向语种切换以及词汇图谱的多语言显示。/ Improved SteamVR subtitle controls, bidirectional language switching, and localization in the Vocabulary Graph.
• 优化音色卡片浏览、新增音色弹窗和角色运动，让界面更清晰、响应更自然。/ Refined voice card browsing, the centered Add Voice dialog, and companion movement for clearer, more natural interaction.

修复 / Fixes

• 修复部分旧设置中翻译风格面板为空、已保存的风格未显示为卡片的问题；旧设置会自动更新并保留自定义内容。/ Fixed empty translation style panels and missing saved style cards in some existing setups. Older settings are updated automatically while preserving custom content.
• 修复窄窗口下同时使用麦克风和系统音频时，设备选项与音量条错位或被裁切的问题。/ Fixed misaligned or clipped device selectors and audio meters when using microphone and system audio together in narrow windows.
• 修复 Haidass 中英翻译可能输出无关内容的问题。该轻量模型使用基础单句翻译，不应用翻译风格或对话上下文。/ Fixed Haidass Chinese–English translation producing unrelated content. This lightweight model uses basic sentence translation without translation styles or conversation context.
• 修复部分中文翻译结果可能引发异常的问题。/ Fixed errors that could occur when processing some Chinese translation results.
• 修复手动测试 TTS 时声音过早中断的问题，并改善中文发音处理。/ Fixed speech cutting off during manual TTS testing and improved Chinese pronunciation handling.
• 修复翻译页面说话时音频波形可能保持平直的问题。/ Fixed the audio waveform remaining flat while speaking on the translation page.
• 修复自动语种切换后可能意外切回中英文的问题。/ Fixed unexpected switches back to Chinese or English after automatic language changes.
• 修复关闭词汇领域后，相关词汇仍可能影响识别的问题。/ Fixed disabled vocabulary domains continuing to influence recognition.
• 修复返回欢迎页后角色未重置位置或重新问候的问题，并完善按钮提示与图标显示。/ Fixed the companion failing to reset its position or greet again on returning to the welcome page, and refined button hints and icons.

当前支持 / Supported Features

• SteamVR 游戏内悬浮字幕与视角跟随 / SteamVR in-game subtitles with gaze tracking
• 麦克风、系统音频与指定应用声音输入 / Microphone, system audio, and per-application audio input
• 可视化 Audio Studio 与自定义音频方案 / Visual Audio Studio and custom audio graphs
• 自动双向语音识别与翻译 / Automatic bidirectional speech recognition and translation
• 本地模型与在线服务，多种模型尺寸、显存和语种提示 / Local models and online services, with model size, VRAM, and language guidance
• NVIDIA 与 AMD 显卡上的本地识别和翻译 / Local recognition and translation on NVIDIA and AMD graphics cards
• 翻译风格卡片与自然中文风格 / Translation style cards and a natural Chinese style
• TTS 中心、内置与自定义音色、声音克隆 / TTS Center, built-in and custom voices, and voice cloning
• 对话上下文与专业术语辅助（取决于所选模型） / Conversation context and terminology assistance, depending on the selected model
• 识别与翻译历史、桌面悬浮字幕 / Recognition and translation history, and floating desktop subtitles
• VRChat OSC Chatbox，直接发送与翻译后发送 / VRChat OSC Chatbox with direct and translated text input
• 自定义 Chatbox 格式、页眉、页脚和消息时长 / Custom Chatbox formatting, headers, footers, and message duration
• VRCX 人名识别辅助与词汇图谱领域管理 / VRCX-assisted name recognition and Vocabulary Graph domain management
• 说话人识别与稳定编号 / Speaker recognition with consistent numbering
• 音视频翻译、字幕预览与导出 / Audio and video translation, subtitle previews, and subtitle export
• 会议录音、转写与记录管理 / Meeting recording, transcription, and record management
• 自动更新与用户设置保留 / Automatic updates with user-setting preservation

下载与运行 / Download and Launch

• Windows：下载 XRTranslate-v0.2.11-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.2.11.exe。/ Download XRTranslate-v0.2.11-win-x64.zip, extract it, and run XRTranslate-v0.2.11.exe.
• Linux x86_64：下载 XRTranslate-v0.2.11-linux-x64.zip，解压后进入文件夹运行 ./xrtranslate。系统与指定应用音频采集需要 PulseAudio 或 PipeWire 的 pipewire-pulse，以及 libpulse.so.0。/ Download XRTranslate-v0.2.11-linux-x64.zip, extract it, and run ./xrtranslate from the extracted folder. System and per-application audio capture require PulseAudio or PipeWire's pipewire-pulse, plus libpulse.so.0.

XRTranslate v0.2.11-beta.3

TTS

文本转语音是测试功能，不建议勾选 TTS / TTS is still experimental and is not recommended for general use.

新增 / New

• Prompt Studio 增加“翻译风格”（Translation Style）浮动面板，支持在画布中直接查看、创建、保存与切换自定义提示词风格，并支持右键直接将文本节点绑定为翻译风格 / Added an interactive Translation Style panel in Prompt Studio for viewing, creating, saving, and switching custom prompt styles, with one-click style binding from text nodes.
• 主界面翻译设置区增加直达“翻译风格”（Translation style）与“翻译器麦克风”（Translator microphone）的快捷入口按钮 / Added quick-access buttons on the translation screen for opening Translation Style settings and Translator Microphone routing.
• Audio Studio 画布增加快捷重置音频系统与清除连接的操作菜单，并添加了常用操作手势与快捷键指引提示 / Added quick-action canvas menus in Audio Studio to reset the audio system and clear links, alongside an on-canvas shortcut guide.

变化 / Changes

• SteamVR Overlay 插件设置页面的参数调节滑块更新为现代滑块样式，提供更平滑的滑动调节与数值反馈，并清理废弃样式 / Updated parameter sliders on the SteamVR Overlay plugin settings page to modern sliders for smoother adjustment and value feedback, removing deprecated legacy styles.
• 优化双向自动语言识别与多模式切换时的语言路由匹配，提升切换过程的平稳性 / Refined language routing and source-target matching during bidirectional automatic recognition and mode switching.
• 版本升级至 v0.2.11-beta.3 / Bumped release version to v0.2.11-beta.3.

修复 / Fixes

• 修复在翻译主界面开麦说话时音频可视化波形条未产生波动、保持平直静态直线的显示问题，解耦电平计算与特定页面门控 / Fixed an issue on the translation screen where the audio waveform meter remained flat and static while speaking, decoupling level metering from page-specific gating.
• 修复删除按钮的无障碍标签与悬停提示信息支持 / Improved tooltip descriptions and accessibility labeling for delete buttons.

下载与运行 / Download and Launch

• Windows: 下载 XRTranslate-v0.2.11-beta.3-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.2.11-beta.3.exe 即可。/ Download XRTranslate-v0.2.11-beta.3-win-x64.zip, extract it, and run XRTranslate-v0.2.11-beta.3.exe.

当前支持 / Supported Features

• SteamVR 游戏内 HUD 悬浮字幕与视角跟随 / SteamVR in-game HUD overlay with gaze tracking
• 麦克风、系统音频与指定应用声音输入 / Microphone, system audio, and per-application audio input
• 可视化 Audio Studio 与灵活的音频路由管理 / Visual Audio Studio with dynamic audio routing management
• 自动双向语言识别与翻译（支持保加利亚语及多种主要语种） / Automatic bidirectional speech recognition and translation (including Bulgarian and major languages)
• Qwen Audio 实时流式语音识别 / Qwen Audio realtime streaming speech recognition
• 本地模型与在线 API 服务 / Local models and online API services
• OpenAI 及兼容接口的语音识别与翻译 / Speech recognition and translation through OpenAI and compatible APIs
• OpenVoice 本地语音合成与声音克隆 / Local speech synthesis and voice cloning with OpenVoice
• 可视化 Prompt Studio、动态分支与自定义翻译规则 / Visual Prompt Studio with dynamic branching and custom translation rules
• 动态上下文、词汇偏置、专业术语与连续对话辅助 / Dynamic context, vocabulary bias, terminology, and conversation assistance
• 翻译异常检测与自动重新生成 / Translation anomaly detection and automatic regeneration
• 实时与离线任务智能推理调度 / Intelligent scheduling for realtime and offline inference
• 识别和翻译历史记录 / Recognition and translation history
• 桌面悬浮字幕 / Desktop floating subtitles
• VRChat OSC Chatbox / VRChat OSC Chatbox
• 在VRCX开启时提高人名识别准确性 / Improve the accuracy of name recognition when VRCX is running.
• Chatbox 直接发送与翻译后发送模式 / Direct and translated Chatbox input modes
• 自定义 Chatbox 格式、页眉、页脚和消息时长 / Custom Chatbox format, header, footer, and message lifetime
• 说话人识别与稳定编号 / Speaker recognition with stable numbering
• 视频和音频媒体识别翻译 / Video and audio media transcription and translation
• 实时生成、预览及导出视频字幕 / Realtime generation, preview, and export of video subtitles
• 会议录音、转写与会议记录管理 / Meeting recording, transcription, and meeting record management
• 自动更新与用户设置保留 / Automatic updates with user-setting preservation
• Linux 系统声音与指定应用声音捕获 / Linux system and per-application audio capture
• 词汇图谱领域管理与可选 VRCX 动态语料 / Vocabulary graph domain management and optional VRCX dynamic corpus
• 不同尺寸的本地识别与翻译模型、显存估计和语种提示 / Local ASR and translation models across size tiers, with VRAM estimates and language guidance

XRTranslate v0.2.11-beta.2

TTS

文本转语音是测试功能，不建议勾选 TTS / TTS is still experimental and is not recommended for general use.

修复 / Fixes

• 修复在语料库中关闭领域后，动态提供源（如 VRCX 运行时）仍残留“VRChat”和“VRC”热词偏置的问题；建立动态提供源 ID（Provider ID）、层级子领域（Subdomain）及前缀的全维度阻断过滤机制 / Fixed an issue where disabling domains in the corpus still retained "VRChat" and "VRC" hotword bias from dynamic runtime providers (such as VRCX); implemented full-dimension filtering across provider IDs, hierarchical subdomains, and ID prefixes.
• 修复双向自动语种选择模式在动态切换到第三方语种后，可能在无输入时意外发生中英文回归的问题，优化了基准语言对保持与状态转移机制 / Fixed an issue in bidirectional automatic language mode where dynamically switching to a third language could trigger unexpected regressions to English or Chinese, refining base-pair preservation and state transition logic.
• 修复控制台与数据管道运行过程中的异常日志冗余输出，精简运行时错误代谢与告警过滤 / Cleaned up redundant error outputs in console and pipeline logs, streamlining runtime diagnostics and logging.

变化 / Changes

• 补全语料库工作室（Corpus Studio）全部内置系统领域与分类的完整国际化多语言支持（中文、日文、韩文、俄文、英文），解决此前“Virtual Worlds（虚拟世界）”等默认根领域显示未本地化的问题 / Added complete multilingual localization for all built-in system domains and groups in Corpus Studio (Chinese, Japanese, Korean, Russian, English), resolving untranslated default root domains like "Virtual Worlds".
• 版本升级至 v0.2.11-beta.2 / Bumped release version to v0.2.11-beta.2.

下载与运行 / Download and Launch

• Windows: 下载 XRTranslate-v0.2.11-beta.2-win-x64.zip，解压后运行文件夹中的 XRTranslate-v0.2.11-beta.2.exe 即可。/ Download XRTranslate-v0.2.11-beta.2-win-x64.zip, extract it, and run XRTranslate-v0.2.11-beta.2.exe.

当前支持 / Supported Features

• SteamVR 游戏内 HUD 悬浮字幕与视角跟随 / SteamVR in-game HUD overlay with gaze tracking
• 麦克风、系统音频与指定应用声音输入 / Microphone, system audio, and per-application audio input
• 可视化 Audio Studio 与灵活的音频路由管理 / Visual Audio Studio with dynamic audio routing management
• 自动双向语言识别与翻译（支持保加利亚语及多种主要语种） / Automatic bidirectional speech recognition and translation (including Bulgarian and major languages)
• Qwen Audio 实时流式语音识别 / Qwen Audio realtime streaming speech recognition
• 本地模型与在线 API 服务 / Local models and online API services
• OpenAI 及兼容接口的语音识别与翻译 / Speech recognition and translation through OpenAI and compatible APIs
• OpenVoice 本地语音合成与声音克隆 / Local speech synthesis and voice cloning with OpenVoice
• 可视化 Prompt Studio、动态分支与自定义翻译规则 / Visual Prompt Studio with dynamic branching and custom translation rules
• 动态上下文、词汇偏置、专业术语与连续对话辅助 / Dynamic context, vocabulary bias, terminology, and conversation assistance
• 翻译异常检测与自动重新生成 / Translation anomaly detection and automatic regeneration
• 实时与离线任务智能推理调度 / Intelligent scheduling for realtime and offline inference
• 识别和翻译历史记录 / Recognition and translation history
• 桌面悬浮字幕 / Desktop floating subtitles
• VRChat OSC Chatbox / VRChat OSC Chatbox
• 在VRCX开启时提高人名识别准确性 / Improve the accuracy of name recognition when VRCX is running.
• Chatbox 直接发送与翻译后发送模式 / Direct and translated Chatbox input modes
• 自定义 Chatbox 格式、页眉、页脚和消息时长 / Custom Chatbox format, header, footer, and message lifetime
• 说话人识别与稳定编号 / Speaker recognition with stable numbering
• 视频和音频媒体识别翻译 / Video and audio media transcription and translation
• 实时生成、预览及导出视频字幕 / Realtime generation, preview, and export of video subtitles
• 会议录音、转写与会议记录管理 / Meeting recording, transcription, and meeting record management
• 自动更新与用户设置保留 / Automatic updates with user-setting preservation
• Linux 系统声音与指定应用声音捕获 / Linux system and per-application audio capture
• 词汇图谱领域管理与可选 VRCX 动态语料 / Vocabulary graph domain management and optional VRCX dynamic corpus
• 不同尺寸的本地识别与翻译模型、显存估计和语种提示 / Local ASR and translation models across size tiers, with VRAM estimates and language guidance