引导式测验 · 中等 等级

聋人测验中的人工智能实时字幕

查看流式语音识别、CART、修订、WER、二值化、字幕放置和字幕可访问性。检查限制和证据。

相关引导路径人工智能为聋人提供实时字幕
问题 1 的 8

什么核心技术将实时语音转换为屏幕文本？

预览每个问题8 问题

当您参加测验时，就会出现答案和解释。

  1. 什么核心技术将实时语音转换为屏幕文本？
    • 自动语音识别 (ASR)
    • 光学字符识别 (OCR)
    • 文本转语音 (TTS)
    • 全球定位（GPS）
  2. What distinguishes CART captioning from automatic speech-recognition captions?
    • It requires no human input
    • A trained human captioner uses stenotype, a computer, and realtime software
    • It translates every caption into another language automatically
    • It is a word-error-rate formula
  3. Why can words in a streaming caption change before an utterance is finalized?
    • The caption file has become an image
    • A speaker label always changes the language
    • The recognizer can emit intermediate hypotheses that change as more audio is processed
    • The final transcript is guaranteed to contain no errors
  4. 通常使用什么指标来衡量字幕准确性？
    • 每秒帧数
    • 分贝
    • 每英寸像素数
    • 字错误率 (WER)
  5. What does speaker diarization add to a live transcription workflow?
    • Labels that distinguish speech from different speakers, without verifying their real-world identities
    • Automatic translation into every language
    • A guarantee of correct word recognition
    • A replacement for microphone audio
  6. What caption capability did Meta describe for Ray-Ban Display glasses?
    • Guaranteeing perfect recognition in noisy rooms
    • Displaying captions for speech directed at the wearer
    • Replacing caption standards for every video
    • Automatically identifying every speaker in a meeting
  7. What do captions for accessibility often include besides spoken dialogue?
    • Only the video title
    • A guarantee of perfect transcription
    • Relevant non-speech audio information such as sound effects or speaker cues
    • 仅模型的置信度得分
  8. Which caption file format does W3C identify as the most common on the web?
    • CSV
    • PNG
    • MP3
    • WebVTT