情感语音合成测验

当您参加测验时，就会出现答案和解释。

1. 情感语音合成主要改变生成音频的哪个方面？
   • 言语的表达方式，例如音高、语速和语音质量
   • 所说的字面意思
   • 音频的文件格式
   • 文本的语言
2. In Russell’s circumplex model of affect, what do valence and arousal represent?
   • Speaker identity and loudness
   • Pleasantness and activation
   • Speech rate and pause duration
   • Articulation and pitch range
3. What acoustic pattern did one study observe in an actress’s deliberately portrayed sad speech?
   • Higher RMS energy and fewer pauses
   • No change in any acoustic measure
   • Lower median RMS energy and more inter-word silence
   • A fixed pitch identical to neutral speech
4. 分类情感模型通常如何告诉合成器使用哪种情感？
   • 通过使用表情符号重写输入文本
   • 通过在生成音频后减慢音频速度
   • 通过更改麦克风设置
   • 通过为选定的情感标签提供学习的嵌入
5. What difficulty did a controllable emotion-transfer paper identify when using reference audio?
   • Selecting a reference that conveys the desired emotion strength
   • Choosing a file format that contains no audio
   • Converting every label into a speaker name
   • Ensuring that all emotion categories use identical pitch
6. For covered providers generating synthetic audio, what transparency measure does Article 50 require?
   • Publishing the source code of every model
   • Marking outputs in a machine-readable and detectable format as artificially generated or manipulated
   • Obtaining an emotion label from each listener
   • Replacing all audio with text captions
7. In reference-based emotion transfer, what information can an emotion embedding carry?
   • A speaker’s written transcript only
   • A guarantee about the listener’s emotional response
   • A learned representation of affective style from the reference speech
   • A replacement for the acoustic waveform
8. What is one goal of cross-speaker emotion disentanglement?
   • Change every word while preserving pitch only
   • Copy the reference speaker’s identity exactly
   • Remove all variation from the generated voice
   • Transfer emotional expression while retaining the target speaker’s identity