這是歷史版本頁面 — 看 NVDA 最新版本

視障者用電腦卡關?NVDA 免費讀屏能做什麼

NVDA 軟體圖示 NVDA 2016.1 資料更新 2016-02-22 閱讀約 6 分鐘

視覺障礙者或重度視力衰退的人,坐在電腦前不是「看不到螢幕」這麼簡單——是滑鼠游標不知道在哪、視窗開了幾個搞不清楚、網頁讀到哪一段完全沒概念。明明文書、上網、收信都是日常需求,卻因為讀不到畫面,每一步都卡在「現在到底發生什麼事」。這種狀況下,需要的不是放大鏡,而是把螢幕上的文字變成聲音或點字,讓操作者用聽的、用摸的就知道系統狀態。

一般人直覺會想到「把字放大」或「請人幫忙唸」,但前者對全盲沒用,後者不可能隨時在旁邊。系統內建的朗讀功能往往只覆蓋特定程式,遇到網頁或 Office 就斷斷續續。這類工具的切入點,是直接攔截系統和應用程式送出的文字資訊,再用語音合成或點字顯示器輸出,讓使用者在沒有視覺輔助的狀態下,照樣能操作 Windows、瀏覽網頁、編輯文件。

NVDA 操作介面
NVDA 的操作介面

01NVDA 是什麼?裝了之後到底能做什麼?

NVDA 是一套免費、開源的螢幕閱讀器,安裝後會用合成語音讀出螢幕上的文字,也能搭配點字顯示器輸出。它讓視障者可以操作 Windows 桌面、瀏覽網頁、使用 Microsoft Office、記事本,以及 Outlook、Thunderbird 等郵件軟體。啟動方式是按下 Ctrl + Alt + N,之後用 Tab 和方向鍵就能在視窗和選單之間移動,系統會即時唸出焦點所在的位置。

02它能不能讀網頁?用鍵盤要怎麼瀏覽?

可以。NVDA 支援 Chrome、Firefox、Edge 等主流瀏覽器,開啟網頁後,用方向鍵上下移動就能逐行讀出文字,按 H 鍵跳到下一個標題,按 K 鍵跳到連結。不需要滑鼠,整頁內容都可以靠鍵盤讀完。遇到圖片或按鈕,它會唸出替代文字或「按鈕」之類的提示,讓你知道這個元素可以做什麼。

NVDA 介面畫面 2
NVDA 介面畫面

03可以接點字顯示器嗎?要怎麼設定?

可以。NVDA 支援多種可更新式點字顯示器,透過 USB 或藍牙連線,然後在 NVDA 的設定選單裡找到「點字」項目,選擇你的顯示器型號並調整相關參數。設定完成後,螢幕上的文字會即時轉成點字輸出,適合不想一直聽語音、或需要精確確認拼字的使用者。

04它一定要安裝才能用嗎?能不能帶著走?

不必安裝。NVDA 提供可攜版,可以把程式直接放到 USB 隨身碟裡,插到任何一台 Windows 電腦上就能執行,不需要在該台電腦上安裝軟體。這對需要在不同地方使用電腦的人很方便,例如在學校、圖書館或別人的電腦上,插入隨身碟啟動 NVDA 就能開始操作。

NVDA 介面畫面 3
NVDA 介面畫面

05它支援哪些 Windows 版本?對電腦硬體有要求嗎?

官方說明支援 Windows 7、8、10、11,32 位元和 64 位元都行。硬體需求是處理器 1 GHz 以上、建議 2 GB 記憶體、100 MB 可用磁碟空間,並需要音效輸出裝置來播放語音。點字顯示器是選配,沒有也可以只用語音。整體來說,一般文書用的電腦都跑得動。

06完全看不到的人適合用嗎?還是只有弱視的人能用?

全盲使用者是 NVDA 的主要對象,因為它完全靠語音和點字輸出,不需要任何視覺回饋。弱視的人也可以搭配螢幕放大鏡工具一起使用,NVDA 本身有與放大工具整合的能力,讓還有一點殘餘視力的人同時用聽的和看的來操作。兩種狀況都適用,差別在於怎麼搭配其他輔具。

NVDA 介面畫面 4
NVDA 介面畫面

07不想花錢買商業讀屏軟體的人,用 NVDA 夠嗎?

如果你的需求是讀網頁、用 Office、收發郵件、操作 Windows 基本功能,NVDA 都能勝任,而且完全免費。它更新頻繁,近兩年有多次版本釋出,代表持續在維護。但如果你需要的是商業軟體才有的進階功能,例如更細緻的語音品質或特定專業應用程式的深度整合,NVDA 可能就沒有那些東西。

08它的介面是英文的,不會英文的人能操作嗎?

官方介面語系是英文,但 NVDA 本身支援多種語言的語音輸出,也就是說,它可以用中文或其他語言朗讀螢幕內容。不過,設定選單、快速鍵說明這些操作介面是英文,對完全不懂英文的人來說,初始設定會比較吃力。建議找懂英文的人協助完成第一次設定,之後日常使用主要靠鍵盤指令,不太需要一直看選單。

09有些程式它讀不到,遇到這種情況怎麼辦?

官方說明指出,NVDA 支援大多數標準應用程式,但某些專有或客製化軟體可能功能有限。如果你常用的程式讀不到內容,可以先確認該程式是否有無障礙支援或替代方案,或者改用網頁版。NVDA 本身沒有辦法解決所有應用程式的相容問題,這部分需要自行測試或尋找其他輔助工具搭配。

10它的語音聽起來很機械,會不會影響長時間使用?

官方說明坦承,NVDA 的語音合成品質比商業付費方案粗糙,聽起來確實比較機械。對長時間聆聽的人來說,這可能造成疲勞,但很多人會透過調整語速、音調和詳細程度來改善。如果你對語音品質非常敏感,可以考慮安裝其他語音引擎,但這需要額外設定,而且效果因人而異。

11跟其他讀屏軟體比起來,NVDA 的評價如何?

在收錄的評分中,NVDA 有 168 人評分,平均 8.18 分(滿分 10 分)。這個分數不算低,但評分人數不算多,參考價值有限。跟它放在一起比較的 Foxit Reader、Adobe Digital Editions、ALLPlayer 都不是讀屏軟體,類別不同,無法直接對比。如果你在考慮付費讀屏,建議先試用 NVDA,確認基本需求能不能滿足,再決定要不要花錢。

下載與官方連結

前往官方網站 下載 NVDA

nvda_2016.1.exe · 17.76 MB · 連結指向軟體原廠

版本紀錄

再看 5 筆

看完整版本清單

這一版改了什麼

What's new in this version:

- Highlights of this release include the ability to optionally lower the volume of other sounds; improvements to braille output and braille display support; several significant fixes to Microsoft Office support; and fixes to browse mode in iTunes.

New Features:
- New braille translation tables: Polish 8 dot computer braille, Mongolian
- You can turn off the braille cursor and change its shape using the new Show cursor and Cursor shape options in the Braille Settings dialog
- NVDA can now connect to a HIMS Smart Beetle braille display via Bluetooth
- NVDA can optionally lower the volume of other sounds when installed on Windows 8 and later. This can be configured using the Audio ducking mode option in the NVDA Synthesizer dialog or by pressing NVDA+shift+d
- Support for the APH Refreshabraille in HID mode and the Baum VarioUltra and Pronto! when connected via USB
- Support for HumanWare Brailliant BI/B braille displays when the protocol is set to OpenBraille

Changes:
- Reporting of emphasis is now disabled by default
- In the Elements List dialog in Microsoft Excel, the shortcut for Formulas has been changed to alt+r so that it is different to the shortcut for the Filter field
- Updated liblouis braille translator to 2.6.5
- The word "text" is no longer reported when moving the focus or review cursor to text objects

Bug Fixes:
- In iTunes 12, browse mode now updates correctly when a new page loads in the iTunes Store
- In Internet Explorer and other MSHTML controls, moving to specific heading levels with single letter navigation now behaves as expected when the level of a heading is overridden for accessibility purposes (specifically, when aria-level overrides the level of an h tag)
- In Spotify, focus no longer frequently lands on "unknown" objects
- Focus is now restored correctly when returning to Spotify from another application
- When toggling between browse mode and focus mode, the mode is reported in braille as well as speech
- The Start buttn on the Taskbar is no longer reported as a list and/or as selected in some versions of Windows
- Messages such as "inserted" are no longer reported when composing messages in Microsoft Outlook
- When using a braille display and text is selected on the current line (e.g. when searching in a text editor for text which occurs on the same line), the braille display will be scrolled if appropriate
- NVDA no longer silently exits when closing a Windows command console with alt+f4 in Windows 10
- In the Elements List in browse mode, when you change the type of element, the Filter by field is now cleared
- In editable text in Mozilla applications, moving the mouse again reads the appropriate line, word, etc. as expected instead of the entire content
- When moving the mouse in editable text in Mozilla applications, reading no longer stops at elements such as links within the word or line being read
- In Internet Explorer, the shoprite.com website can now be read in browse mode instead of reporting as blank. (Specifically, malformed lang attributes are now handled gracefully)
- In Microsoft Word, tracked changes such as "inserted" are no longer reported when track changes markup is not displayed
- When a toggle button is focused, NVDA now reports when it is changed from pressed to not pressed
- Reporting of mouse shape changes again works as expected
- When speaking line indentation, non-breaking spaces are now treated as normal spaces. Previously, this could cause announcements such as "space space space" instead of "3 space"
- When closing a modern Microsoft input method candidate list, focus is correctly restored to either the input composition or the underlying document
- In Microsoft Office 2013 and later, when the ribbon is set to show only tabs, items in the ribbon are again reported as expected when a tab is activated
- Fixes and improvements to touch screen gesture detection and binding
- Touch screen hovers are no longer reported in input help
- NVDA no longer fails to list comments in the Elements List for Microsoft Excel if a comment is on a merged cell
- In a very rare case, NVDA no longer fails to read sheet content in Microsoft Excel with reporting of row and column headers enabled
- In Google Chrome, navigating within an Input composition when entering east Asian characters now works as expected
- When searching Apple Music in iTunes, browse mode for the search results document is now updated as expected
- In Microsoft Excel, pressing shift+f11 to create a new sheet now reports your new position instead of reporting nothing
- Fixed problems with braille display output when entering Korean characters

Changes for Developers:
- The new audioDucking.AudioDucker class allows code which outputs audio to indicate when background audio should be ducked
- nvwave.WavePlayer's constructor now has a wantDucking keyword argument which specifies whether background audio should be ducked while audio is playing
- When this is enabled (which is the default), it is essential that WavePlayer.idle be called when appropriate

Enhanced I/O for braille displays:
- Thread-safe braille display drivers can declare themselves as such using the BrailleDisplayDriver.isThreadSafe attribute. A driver must be thread-safe to benefit from the following features
- Data is written to thread-safe braille display drivers in the background, thus improving performance
- hwIo.Serial extends pyserial to call a callable when data is received instead of drivers having to poll
- hwIo.Hid provides support for braille displays communicating via USB HID
- hwPortUtils and hwIo can optionally provide detailed debug logging, including devices found and all data sent and received

There are several new properties accessible from touch screen gestures:
- MultitouchTracker objects now contain a childTrackers property which contains the MultiTouchTrackers the tracker was composed of. For example, 2 finger double tap has child trackers for two 2-finger taps. The 2-finger taps themselves have child trackers for two taps
- MultiTouchTracker objects now also contain a rawSingleTouchTracker property if the tracker was the result of one single finger doing a tap, flick or hover. The SingleTouchTracker allows access to the underlying ID assigned to the finger by the operating system and whether or not the finger is still in contact at the current time
- TouchInputGestures now have x and y properties, removing the need to access the tracker for trivial cases
- TouchInputGesturs now contain a preheldTracker property, which is a MultitouchTracker object representing the other fingers held while this action was being performed

Two new touch screen gestures can be emitted:
- Plural tap and holds (e.g. double tap and hold)
- A generalized identifier with finger count removed for holds (e.g. hold+hover for 1finger_hold+hover)