本次更新改善了语音的清晰度、编辑时的时间精度和长项目的导出可靠性，让从语音生成到视频制作的操作更加顺畅。对于需要处理长音频、多个视频片段或精确字幕时间的项目，也带来了更一致的编辑体验。 • 改善了 Supertonic 语音的发音和重音表现，减少生成语音时遗漏单词的问题。七种 VoxCPM2 土耳其语音继续共用同一个模型包。这个包的下载恢复更加可靠，意外中断后能够更稳定地继续完成，减少重复尝试下载却仍未完成的问题。 • 音频与视频互相转换后，项目现在会显示转换后的正确媒体类型，便于确认当前内容。将长音频转换为视频时会保留完整时长，也包括长达两小时的录音，避免生成的视频提前结束。需要将较长的节目制作成视频时，可以更可靠地保留结尾内容。 • 改善了长达一小时的照片项目的导出可靠性。照片会一直保留到项目结束，方便为较长的音频内容制作具有完整时长的画面。为一小时左右的节目准备照片画面时，照片的显示时间也会与项目结尾保持一致。 • 人脸遮挡和效果工具的时间选择更加一致，便于设置效果出现和结束的位置。Magic Trim 支持精确到百分之一秒的调整，让细小的剪辑修改更容易控制。 • 调整音频淡入、淡出时长时，VoiceOver 焦点更加稳定，操作过程中更容易继续编辑。撤销和重做按钮也更加紧凑，减少对编辑界面空间的占用。 • 改善了五种动物和昆虫的音效，使其听起来更加自然，也更适合直接用于项目中的声音设计。 • 当项目无法使用字幕或音频分离功能时，现在会更清楚地说明原因，帮助您了解当前媒体内容可以使用哪些工具。这些说明让功能暂时不可用时的状态更容易理解，也便于安排接下来的编辑步骤。 • 合并 4K 和可变帧率片段、裁剪视频或改变速度时，更好地保持画面的流畅度以及声音与画面的同步，同时保留 HDR 色彩和视频方向，减少连续编辑带来的时间偏差。对于同一项目中先合并、再裁剪或调整速度的操作，声音和画面的对应位置也更加可靠。 • 音频描述的时间定位更加精确，有助于将描述放在需要的位置，让描述的开始和结束更贴合项目中的目标时间。修正了错误提示描述超出项目结尾的问题，避免有效内容被不准确的警告打断。 • 在口述影像中，如果 SRT 或 VTT 文件超出视频时长，您可以在导入前查看并确认调整。确认后，将跳过在视频结束时或之后开始的条目；对于跨越视频结尾的条目，仅将结束时间调整到视频结尾。保留条目的文本和原始 SRT/VTT 文件保持不变。

错误修复

• 改进了 VoxCPM2 语音模型的下载进度显示、中断后续传和错误提示。 • 修复了已保存的 MP3 项目被误认为视频、要求转换为音频的问题。 • “移除内嵌字幕”快捷按钮现位于“字幕”旁边，样式与其他工具栏按钮一致。 • 改进了保存项目时的文件保护。

• Premium 声音库新增 7 种 VoxCPM2 土耳其语声音，改进了声音示例、文本转语音和播放。内置示例无需下载大型语音模型即可收听。 • 新增“移除内嵌字幕”：选择检测到的文字和时间范围，或使用无障碍控件手动添加区域。可选下载约 40 MB 模型后，即可离线处理，无需上传视频。支持前后对比、保留视频音频以及撤销更改。复杂画面可能留下痕迹，处理时间也可能超过视频时长。 • 在 Premium > 已下载的模型中查看模型包名称和存储占用，删除不需要的模型，并在以后重新下载。 • “添加音频”现可使用已保存的音频项目，也可在您确认后将视频项目转换为音频。修复了影响 iPhone 上 MP3 分享的文件处理问题。

错误修复。

通过本次更新，你可以使用人工智能创作音乐。立即前往“更多工具”探索这项功能。免费、无限的 AI 音乐……此外，我们还修复了许多在测试中发现的问题，以及各位尊贵用户通过反馈报告的问题。

• 新增面部遮挡工具：使用模糊、马赛克或黑色遮罩，遮挡所有检测到的人脸或选定的人脸片段。 • 可选择 Apple Vision、开源 MediaPipe 或两者结合，在设备上检测人脸。支持添加手动遮罩、调整移动方式和生效时间，并在应用前预览结果。 • 改善了大字体下时长、开始和结束字段的显示。移除了棕褐色和像素化页面中的空白按钮，并修正了 Premium 代码部分的翻译。 面部遮挡无法保证所有人脸都被完整遮住。分享前，建议请视力正常的人士完整观看并检查视频。

版本 5.1 新增 • AI 放大：使用完全在设备上运行的 AI 模型，让视频更清晰并放大。 • AI 提升帧率：生成中间帧让动作更流畅，提供 90 和 120 帧/秒选项。 • 分轨分离：把音频或视频的声音分成人声、鼓、贝斯和其他声音；可分别调节音量并重新混音。 • 取景助手：通过相机语音提示面部位置、距离、光线和清晰度，还可检查已打开视频的取景。 • 人脸跟踪竖屏取景：跟随人脸把横屏视频转换为 9:16、4:5 或 1:1，并提供模糊背景选项。 • Whisper 离线字幕：自动字幕和翻译字幕可选择 Apple 或 Whisper。模型已内置于应用，无需额外下载。 • 真正的 MP3 导出：音频和视频项目均为 192 kbps。 • DeepFilterNet3 降噪现为默认选项，强度、混合比例和前后对比试听均可调节。 • 无障碍时间轴：排列片段，并添加独立的音乐层和文字层。 • 可迁移项目：导出和导入项目，并接收来自其他应用的媒体。 • 以照片开始：在“新建”中选择照片，并设置时长和效果。 • 我的项目：每个项目是一个独立的 VoiceOver 项目；重命名、导出和删除位于“操作”转子中。 修复 • 时间轴编辑不再把 60 帧/秒的素材降到 30；每个片段保持自己的帧率。 • 90 和 120 帧/秒的导出会保持帧率。由于“照片”可能把高于 60 的文件当作慢动作，无法避免时会显示提示。 • 加载失败的项目可能以上一个文件打开，并在保存时覆盖新项目。已修复。 • AI 工具启动前会检查可用存储空间，其错误现在显示在工具界面上。 • 提升帧率不再在尺寸为奇数或时间戳异常的视频上失败。 • 音频项目中需要视频的按钮会变暗，无声视频中的声音工具也会变暗，并朗读原因。 • 从其他应用打开文件时，不再移动或删除原始文件。 • 来源缺失的幻灯片可能删除旧的资源文件夹。已修复。

zEdit 5.0 来了！这是我们迄今为止最重大的更新之一！时间线让您能够更细致地编辑项目。新增单声道和立体声音频设置、MP3 支持、标记功能和高级降噪。现在还可以向项目添加照片。您可以导出项目并与其他 zEdit 用户分享。全新 Premium 功能，让 zEdit 更智能！配音和口述影像语音：24 种声音，支持 31 种语言。全新本地 AI 功能：离线获取视频和图像描述、提问，并进一步了解您不满意的内容。让您更轻松地按自己的意愿进行编辑！*AI 功能在您的设备上运行，您的数据绝不会离开设备。

Bug fixes.

Bug fixes.

Bug fixes.

版本 3.8 新增 • 图片叠加的入场效果。提供二十种效果：自左侧与右侧滑入、落下、升起、放大、缩小、旋转、横向与纵向展开、幕布、百叶、马赛克、雨落、扫过、对角、涟漪、拉链、弹跳与淡入。未选择效果时，行为与此前版本一致。 • 音频滤镜强度调节。可调整每个滤镜的强度，试听按所选强度生成。已添加的滤镜可事后编辑，更新其时间区间与强度。 • 开场标题在添加时即刻处理。该操作与撤销、重做兼容，并可在执行过程中取消；取消后素材保持不变。 修复 • 变调类音频滤镜未保持时长。氦气、巨人与低沉机器人播放偏快，片段末尾无法听到。变调已改用时间伸缩方式重写，时长与速度得以保持。氦气另增加语音频段提升并减小变调幅度；滤镜特征得以保留，语音仍清晰可辨。 • 增加了针对导出失去响应的防护。若进度长时间无变化，操作将中止并给出明确错误提示；此外导出现已可取消。对高帧率素材，叠加处理固定为每秒 60 帧，不影响时长与速度。 • 视觉效果菜单中的时长选择改为使用小时、分钟与秒字段，启用毫秒设置时另提供百分之一秒字段。原先的半秒步进控件已移除。

版本 3.7 新增 • 视频图片叠加。从相册或“文件”中选择的图片可在设定的时间区间内或整段视频中显示。保持原有宽高比，竖向图片两侧以黑边填充，图片不会旋转或拉伸。与项目保存和撤销兼容。 • 自动工作保护。编辑会话在后台写入恢复记录。应用意外关闭时，将在下次启动时提示继续该会话。恢复的工作同时列于“我的项目”，显示占用空间，可在此打开或删除。仅保留一条记录，每次覆盖写入，并在一周后自动清除。 • 插入流程扩展。内容可取自相册或“文件”，除视频与音频外现已支持插入照片。照片可设定显示时长。 • 音频滤镜试听改为由源素材生成。自随机位置截取十秒片段，以所选滤镜播放。素材较短时使用全部内容。 • 滤镜处理在添加时立即开始。此前需退出菜单后方才启动。 • 设置中的人工智能部分已重新组织。各服务商归入独立卡片并直接显示连接状态；身份验证、模型与验证分列于独立标题之下。说明文字已精简。 修复 • 修复以低采样率音频源插入静音时的崩溃。输出设置现依据素材在有效范围内确定，若被拒绝则自动改用兼容配置。 • 修复开场标题未在视频中显示、原位留有黑色区域的问题。 • 修复字幕仅显示背景条而不显示文字的问题。 • 修复导致字幕与水印显示于随机片段而非所选区间的计时错误。 • 音频滤镜已重新调校以保持语音清晰度。滤镜特征得以保留，语音内容仍可辨识。 • 修复关闭页面后试听音频继续播放的问题。 • 已从“添加内容”菜单中移除不一致的标题与重复说明。 • 所有时间选择器现均显示分钟字段。启用毫秒设置后，时间值在屏幕与读屏中均以百分之一秒精度呈现。 • 修复照片显示时长文本在七种语言中的错误呈现。 • 降低了长时间操作期间应用被终止的可能性。处理期间屏幕不再自动锁定，导出时每帧内存占用已下降。

版本 3.6 新增 • 导出质量现在可在“设置”中选择。兼容可在任何设备上播放，均衡为默认选项，最高会保留你原片的编码，在相同文件大小下画质明显更好。 • 调整比例时，画面如何摆放由你决定。选好比例后，“适应”保留完整画面，“填充”则彻底去掉黑边；把横向视频改成竖向时，左右两侧会裁掉一部分。 修复 • 修复了从相册打开视频时出现“操作无法完成”的错误。导入现在分三级进行，每一级都会确认文件确实能打开。 • 修复了部分 iPhone 机型后置摄像头拍摄的视频无法打开的问题。文件现在会按其封装格式对应的扩展名保存。 • 修复了导出的视频在相册中被当作慢动作的问题。只有高于 60 的帧率会降到 60；以 30 或 60 帧拍摄的视频不作改动，时长和速度完全保持不变。 • 修复了编辑区“添加音频”页面打开后空白的问题。 • 导出画质明显提升。高帧率视频不再以一半的码率编码，视觉效果步骤不再把声音降为单声道，音频采样率也与源文件保持一致。 • 水印、字幕和开场标题合并为一次处理。三次编码变成一次：画质更好，导出也明显更快。

VERSION 3.5 WHAT IS NEW • AI support. Enter your Gemini, OpenAI, Grok or Claude API key in the new Artificial Intelligence section of Settings. The key is stored only in your device secure keychain; requests go straight from your own account, the app is not in between and charges nothing. For each provider you can pick a model, refresh the model list from the provider, and test the connection with one tap. • AI Description. In the editor, frames are taken from the range you choose and the AI writes timed descriptions for blind viewers. You set the density and the style. Read the texts and send the one you like to the Audio Description tool with one tap. • Ask the AI. Ask anything about the part of the video you choose. Video frames and the spoken text, if any, are sent together, so both visual and audible questions are answered. • You can select more than one AI. Each model answer appears in its own section, so nothing is lost inside another answer and you can compare models. • Automatic subtitles can now be produced by AI. On-device recognition is free but makes mistakes; with a key, subtitles are produced with your own subscription and are noticeably more accurate. You choose the method each time. • In audio description, the providers natural voices were added to the voice list. They work in exactly the same flow as device voices. FIXES • Error messages are now really shown in the language you selected. A mapping bug made every processing error appear in English no matter which language was chosen. • The rate slider and the sample button, which do nothing for AI voices, are no longer shown for them.

VERSION 3.3 FIXES • Fixed the app quitting unexpectedly on long videos. Processed frames accumulated in memory until the system terminated the app. Memory now stays flat regardless of video length. The same protection was applied to slideshows, photo-to-video, audio processing and all analysis steps. • Fixed speed changes failing after the first attempt, and added an automatic retry if an export is rejected. • Fixed exported videos having width and height swapped on recordings that carry rotation information. • Ducking the video's sound during audio description now works reliably; "mute completely" and zero percent had no effect on some videos. • Precise position control with the rotor now works. NEW • With hundredths shown, "forward/back one hundredth" buttons appear under the time pickers for one-tap fine tuning. • Volume levels in audio description adjust in five-percent steps, with VoiceOver too. • Loop Forever: added audio repeats until your media ends. • Auto-ducking: music is gently lowered during speech. • Enhance Voice: one-tap noise reduction. • More Options: export the selected part as a separate file or turn it into a new project.

VERSION 3.2 The star of this release: projects can now be added to each other. NEW • More Options: in Cut Between Two Points and Magic Trim, the part you selected can now be exported as a separate file or turned into a new project with a unique name — without touching your open media. Exports go through the app's quality- and compatibility-guaranteed shared export path. • Loop Forever: when adding music, a new "Loop Forever" option with a "more info" button next to it. When on, the added audio repeats from the beginning until your media ends — a short piece of music plays until the video finishes. The fade-out lands at the end of your media, and auto-ducking works across all repeats. • Auto-ducking: when adding music, a new "Auto-duck music during speech" option finds the speech in your video with energy analysis and gently lowers the music during it — the classic radio and documentary mixing rule, in one tap. • Enhance Voice: one-tap strong noise reduction. It suppresses outdoor rumble and wind and brings speech forward; the result can be undone. • Precise position control with the rotor: the player's "current position" is now a VoiceOver-adjustable element. Swiping up or down moves the position second by second, with haptic feedback on each step if your vibration setting is on. • "Choose from Projects" comes to Add to Beginning / End. Pick one of your saved projects; every pending edit in it — subtitles, sound effects, audio filters, watermark, visual effects and audio descriptions — is processed automatically, step by step, and the result is added to the beginning or end of your open media. Progress is announced with percentages throughout. The project you picked is never modified; it stays a project. • Type matching is handled intelligently: while editing a video, audio projects (and vice versa) are listed under "Projects of a different type" and require your confirmation. If you confirm, an audio file is added as a black-screen video with its sound; a video is added with its picture removed, audio only. FIXES • Precise position control with the rotor now works: swiping up or down on the player's "current position" element with VoiceOver had no effect. The element is now genuinely adjustable and swiping moves the position second by second. • In audio description, the spoken duration of very long texts is now measured correctly: the measurement's safety limit was a fixed 20 seconds and cut long texts short, making the duration info and its overflow warning wrong. The limit now scales with the length of the text.

VERSION 3.1 A speed and precision update. IMPROVEMENTS • On launch, leftover temporary files from interrupted operations are now cleaned up automatically, so they no longer take up space on your device. • The frozen frame used when inserting silence is now derived from the source's actual frame rate, making it reliable on 24 fps and variable-frame-rate recordings too. • Exports and other long operations are now protected in the background: briefly switching to another app no longer interrupts them. For very long operations that exceed the system limit, the work resumes from where it left off when you return. • Capture date, location and title of processed media are now preserved: exported videos and audio land in your library at their original date instead of today's. • Adding content to the beginning/end is now many times faster: technically compatible videos (same resolution and format) are joined without re-encoding, preserving quality exactly. Content with a different size or orientation still goes through the safe path with correct fitting. FIXES • The "Use current position" buttons now write exactly the second you see on screen; being at second 30 and getting 31 is fixed. Every position button in the app (trim, range trim, insert content, insert silence, audio filters) now follows the same rule. • Fixed being able to pick a time longer than the media in the trim screens: values beyond the end are automatically clamped to the end of the media. • The warning shown before cutting or inserting content now tells the truth: it can be triggered not only by subtitles but also by pending sound effects, audio filters, visual effects or a timed watermark; the text now lists all of them and notes that removed edits can be recovered with Undo. • Prevented possible edge artifacts when a re-encoded video's resolution landed on an odd number.

VERSION 3.0 This release focuses on picture quality and audio control. FIXES • Fixed portrait photos appearing sideways in multi-photo slideshows and in add-music-to-photo: the photo's orientation is now applied. • Fixed the picture turning black after adding audio to a video: during fast processing the video's format information wasn't being carried into the output. Also fixed ordinary wide-gamut videos being mistaken for HDR. • Fixed colors shifting, the picture washing out or distorting in videos with sound effects, filters or added music. Color properties are now preserved exactly from the source, and HDR (Dolby Vision) recordings are correctly detected and processed as HDR. • Fixed Apply and Preview staying dimmed from the start in the trim screens: the end time now defaults to the end of the media, so you can trim as soon as the screen opens. In Magic Trim, marking only the start is enough. • The mute option in the add-audio screen was renamed to say exactly what it does. • Visual effects now appear in the preview too; previously they only showed up in the exported video. • Fixed the permission error when saving projects on iPad. • Fixed the share sheet offering no options for some audio formats. NEW • Insert Content: you can now insert video (not just audio) into the middle of a video. Adding to the beginning/end lives under the same button. • "Use current position" in the trim screens, and the same convenience when inserting silence. • Your selections are kept when returning from the trim preview.

VERSION 2.9 This update focuses on a comprehensive stability review. NEW • The "Insert" feature is now called "Insert Content" and does more: into a video you can insert not only audio but also video (for example a meme) in the middle. Adding to the beginning/end is now an option under the same "Insert Content" button. • A "Use current position" button was added to the trim screens: pick the player's current moment for start and end with one tap. FIXES • Fixed the "you don't have permission" error when saving projects on iPad (and in some cases on iPhone): the projects folder is now created reliably, file copying was made resilient, and a clear message is shown if anything fails. • When you Cancel from the trim preview, your chosen start and end times are now kept, so you can adjust and retry without entering them again. • Fixed centisecond pickers in time fields sometimes not reflecting the chosen value (notably the start time not responding when moved backwards): time is now computed with integer arithmetic. • Fixed the share sheet offering no sharing options (only saving) for some audio formats (WAV, AIFF): the file type is now declared explicitly to the system. The format picker also explains which format is compatible with every app. • Every video operation (trim, split, mute, add audio, sound effect, filter, insert silence) now goes through a single, safe export path. This prevents black video, video turning into audio, or corrupted output on iPhone HDR (Dolby Vision) recordings at the root: the app verifies its own output and rebuilds it with the safest method when needed. • Sharing was made safe everywhere in the app; an empty or broken file is caught before sharing, preventing the "cannot be shared" error. • Video quality and resolution are preserved in every operation. The improvements were reviewed across the whole app as a coherent whole.

VERSION 2.8 This update focuses on stability and audio tools. FIXES • Fixed silence trimming turning video into audio or producing a black picture, including on iPhone HDR (Dolby Vision) recordings. Video always stays video, rebuilt with the safest method automatically when needed. • Fixed export failing or coming out empty when a sound effect or audio filter is added; an empty file is now caught before sharing and a clear message is shown. Videos with sound effects appearing broken in the gallery are also fixed. • Fixed being unable to paste text from the clipboard into the subtitle field. NEW • Audio Compressor: evens out volume differences so whispers are audible and shouts aren't harsh, with an optional Limiter for broadcast-ready sound. In the editor and in Extra Tools. • Fade Out: gently lowers the audio to silence toward the end of the recording, for a professional ending. In videos the picture is untouched. • Insert Silence: adds silence of your chosen length at the position you pick — ideal for leaving a gap after a sound effect before speech begins. In the editor and in Extra Tools. • Custom durations in Smart Split: set each part's length yourself; any remainder goes to the last part. • Transcription now works end-to-end on long recordings.

VERSION 2.7 NEW • Audio Compressor: evens out volume differences — whispers become audible, shouts stop being harsh. Available in the editor and in Extra Tools, with an optional Limiter ceiling guard; decide by listening to before/after. • Advanced Speed Change: apply to the whole media or a chosen range, eight speeds from 0.25x to 4x; adjust while listening to a sample. Pitch and quality are preserved. • 7 new audio filters: three new robots, Helium, Ghost, Stadium Announcer and Giant. Underwater and cave filters were also redesigned; there are three caves now. • Custom durations in Smart Split: instead of equal parts, set each part's length yourself; any remainder goes to the last part, with live duration info guiding you. • Visual effects are now saved with the project and covered by undo. FIXES • Fixed being unable to export after adding audio to a video, and options staying dimmed. • Fixed visual effects not appearing in the preview and export. • Permanently fixed silence trimming turning some videos into audio-only. • Fixed the watermark position: the corner you choose now matches the exported video. • Fixed sharing: exported content now transfers to every app without errors. • Fixed videos playing silently when the phone's silent switch is on. • Transcription now works end-to-end on long recordings: even an hour of audio is recognized piece by piece and fully transcribed. • Fixed exported video coming out completely black in projects made of several joined clips.

VERSION 2.6 WHAT’S NEW • 130 HD sound effects + sound import from royalty-free sources. • Effect volume, playback duration and VoiceOver-friendly duration labels. • Text watermark with position, size, opacity and time controls. • Live Preview for effects, filters, subtitles and watermark. • Smart Tools: Sound Map, Filler Cleanup and Scene Draft. • Auto-save drafts in Audio Description. • Live audio mix preview with smooth music fade. • Scene Draft now supports Photos videos. • Contributors page. • Advanced Speed Change (0.25x–4x) with range support and preserved quality. • 20 cinematic visual effects with live preview. FIXES • Fixed audio description resource leak and empty project issue. • Fixed insert, undo/redo and subtitle/filter warnings. • Fixed Add Audio freeze with large/cloud files. • Fixed silence trimming converting videos into audio. • Improved filler-word detection and multilingual error messages.

zEdit 2.5 — What's New NEW FEATURES • Sound Effects: 139 ready-made effects in 11 categories to place on the timeline — applause, laughter, win/lose stings, percussion, nature, animals, mechanical sounds, cinematic transitions and comedy classics. All effects are generated inside the app and are completely royalty-free. Preview each one before adding, choose the start with hundredth-of-a-second precision; effects are saved with your project and mixed into the audio on export. • Audio Filters: 10 filters that transform a chosen range of your audio — Echo, Cave Echo, Room Reverb, Telephone, Old Radio, Underwater, Megaphone, Robot, Tremolo, Muffled. Start and end are set with accessible fields, every filter can be previewed, and logic errors are reported with clear messages. • Undo and Redo: Actions like trimming and volume are undone with one tap; the history is saved with the project. Better yet, subtitles, effects and filters removed by a time-changing action come back to their lists when you undo that action. • Insert: Place new content right into the middle of your video at the second you choose; existing content is split and continues without interruption. • Smart Video/Audio Split: Divide a file into equal parts, delivered as one numbered ZIP. Video is not re-encoded: fast and lossless. • Receive via Share: Content shared from Voice Memos, Photos or Files greets you straight in the Editing Area. • 4 New Languages: Chinese (Simplified), Korean, Ukrainian and Portuguese — the app is now complete in 13 languages. • Audio Merge renewed: Real file names are shown, files can be reordered with move up/down (including VoiceOver custom actions), each file can be previewed before merging; merging always follows the visible list order. FIXES • The rare data loss in audio description projects is fully eliminated; your work now saves automatically — closing the app can no longer lose a description. • Speech synthesis stalls behind the "Audio could not be created" error are fixed; unedited texts are now genuinely skipped. • The freeze during noise reduction is fixed; processing runs in the background and the interface stays responsive. • Music drifting by a few seconds in photo slideshows is fixed at the root. • Share sheets that could open empty are repaired; volume adjustment is reliable on long recordings; audio overlay and silence trimming are significantly faster. • Many accessibility and stability improvements.