翻唱歌曲识别测验

当您参加测验时，就会出现答案和解释。

1. Why can a recording fingerprint designed for near-identical audio fail on a cover performance?
   • A cover may change tempo, key, arrangement, and instrumentation while preserving the composition
   • It is guaranteed to match any recording with the same title
   • Every cover uses the same waveform samples
   • It can identify a song only from album artwork
2. 色度 (HPCP) 特征代表什么？
   • 每秒的准确分贝级
   • 12 个八度音级折叠在一起
   • 表演音乐家的名字
   • The file’s compression format
3. 系统如何处理以不同音调录制的翻唱？
   • 他们放弃并报告没有匹配
   • 他们将音频转换为文本
   • 他们循环旋转 12-bin 色度向量来测试所有转置
   • 他们将录制速度加快至 2 倍
4. 什么技术可以拉伸一个音频序列以与另一个具有不同节奏的音频序列对齐？
   • JPEG 压缩
   • 随机洗牌
   • 像素插值
   • 动态时间扭曲 (DTW)
5. How can a learned cover-song model support retrieval across tracks of different lengths?
   • Map each track to a fixed-dimensional representation and compare representations
   • Require every version to have the same waveform length
   • Search only by file name
   • Compare the full waveform sample-by-sample without alignment
6. Which training loss does ByteCover use to compact same-composition versions in its embedding space?
   • A reconstruction loss only
   • Triplet loss, alongside a classification loss
   • A binary cross-entropy loss between raw waveforms
   • A next-token prediction loss
7. How can chroma features help compare some cover recordings?
   • They make every instrument produce identical features
   • They remove all effects of arrangement and production
   • They summarize pitch-class structure and reduce sensitivity to octave and some timbral changes
   • They prove two tracks are the same composition
8. What information can a single global chroma summary lose when comparing cover songs?
   • The fact that pitch classes repeat across octaves
   • The recording’s file extension
   • Whether the waveform has a left and right channel
   • The temporal order of pitch-class events through the track