WeChall - DropDaBass

Challenge

题目提供一个 1.12 MB 的 MP3(DJ Anto - Drop Da Bass.mp3,74.08s,44.1kHz,128kbps,joint-stereo),要求找出音乐中隐藏的 message。论坛确认答案是一个 "pronounceable short sentence"。

结论(截至 2026-09-10)

这是出题者设计为「耳朵 + 纸笔」的关卡(gizmore:"only need a special hardware: your ears" / "Maybe listening + pen&paper works best for this one")。歌词与 metadata 提供了操作指令,但隐藏句子本身必须人工听音确认。以下是全部线索与已实测证否的自动解码路线,供后续人工听音时参考,不猜答案。

已知线索

  1. MP3 ID3 tag:

    • TIT3 = record_description.php(指向歌词页)
    • WXXX = Hello World! Nothing more here! ;)
    • COMM = Checksum : 536F20796F75207468696E6B206974277320746861742045415359206875682021213F3F0D0A
    • 十六进制解码:So you think it's that EASY huh !!??(提示,非答案)
  2. 歌词(record_description.php?lyrics=show)主段关键句:

    1
    2
    Drop da bass, Got to listen to the clap
    Come on I will give you the password
    说明:去掉/削弱 bass 后专门听 clap。

  3. 论坛(forum-t303/Some_pointing,Wayback 存档 p1/p2 全文确认):

    • 只需听音频;DSP 是好的搜索词;听 + 纸笔最好
    • "the hidden lyrics give a hint. You have to "drop" the base for best listening."
    • "The lyrics are the instruction. You need to listen closely to extract the message."
    • 答案是一个可发音的短句(gizmore + MrStorm 确认)
  4. 法语注释(intro 段自述): > c'est tellement accéléré qu'on comprend pu rien > (which means in french: It's so accelerated that we don't understand anything)

    配合 (accelerating) / (decelarating) 标注:intro 有一段被加速到不可懂的人声;歌词里的 secret fussoir de touindin1234 tagadap tsouin tsouin 等行就是加速语音的近似听写。

自动分析记录(全部实测,均无有效结果)

  1. 声道检查:L/R corr ≈ 0.976,L-R 差分能量几乎为 0 → 无相位反转 / 侧声道信息。
  2. Whisper large-v3 转写(原曲 / 多种高通带通 / 0.33x–0.66x 变速 / 倒放 / 门限放大 / intro / 主段 / password 段,beam 5,temperature 0/0.6):全部输出 Whisper 经典幻觉模板(Thank you for watching!I don't care if you loseI'll see you in the next video 等),无稳定可懂句。
  3. 180 BPM、8-beat 网格(1/6 s)上做 clap 存在/缺失二值化并按 8-bit 分组:无有效 ASCII;1.15–2.5 阈值扫描 + 移位扫描均无文本。
  4. clap 事件间隔分析:主要打击固定落在 1/6 s 网格(hi-hat 层),低频重音只在小节边界成对出现(~0.33s 间隔、成簇),无 Morse 式长短间隔。
  5. clap 频谱重心 / 对数梅尔 k-means 聚类:重心连续分布(无离散「音高→字母」映射),聚类只反映段落结构。
  6. MP3Stego:编译官方 1.1.19 Decode.exe(wine 运行),空密码及 EASY/easy/完整注释句等候选密码,2836 帧全部处理但报 unexpected end of cipher message,无有效载荷 → 排除。
  7. Spectrogram(多窗口/多频段/高分辨率,含视觉复核):无可读文字、摩斯结构或离散频率组。
  8. binwalk / 文件尾部:无附加文件。

生成听音文件的可复现命令

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
#!/usr/bin/env bash
set -eu
input="${1:-DJ Anto - Drop Da Bass.mp3}"
outdir="${2:-listen}"
mkdir -p "$outdir"

# 去 bass 主段
ffmpeg -y -hide_banner -i "$input" \
-af 'highpass=f=400,loudnorm' -ac 2 \
"$outdir/drop-bass.wav"

# 主段 0.75x 去 bass
ffmpeg -y -hide_banner -ss 24 -t 34 -i "$input" \
-af 'highpass=f=400,loudnorm,atempo=0.75' -ac 1 \
"$outdir/main-drop-bass-slow.wav"

# 主段 0.5x 保留 clap 频段
ffmpeg -y -hide_banner -ss 24 -t 34 -i "$input" \
-af 'highpass=f=2500,lowpass=f=15000,loudnorm,atempo=0.5' -ac 1 \
"$outdir/main-clap-slow.wav"

# 38–56s("password" 唱段)0.5x 保留 clap 频段
ffmpeg -y -hide_banner -ss 38 -t 18 -i "$input" \
-af 'highpass=f=2500,lowpass=f=15000,loudnorm,atempo=0.5' -ac 1 \
"$outdir/clap-38-56-slow.wav"

# intro 0.5x(加速语音提示区)
ffmpeg -y -hide_banner -ss 0 -t 24 -i "$input" \
-af 'atempo=0.5,highpass=f=300,volume=5' -ac 1 -ar 16000 \
"$outdir/intro-slow.wav"

人工听音记录与后续建议

已试听 listen/main-clap-slow.wavlisten/clap-38-56-slow.wav(24–58s 半速 + clap 频段),未听出信息。建议后续按以下顺序尝试:

  1. intro(0–24s)减速听:listen/intro-slow.wav 或自行 atempo=0.5atempo=0.5(0.25x),对照歌词听写行(secret fussoir de touindin 等)找「减速后可懂」的句子;
  2. 主段 38–58s 用耳机、低音量重复听人声层,注意歌词里没有印出的词;
  3. Audacity 频谱图:0–5kHz、窗口 1024–2048,检查中频是否有被音乐盖住的文字/图形(提示里的 DSP 词条与 9gag 时代做法都偏可视化);
  4. 若认为 clap 节奏编码:把记录下来的「每个 beat 上有没有 clap」按 1/6 s 网格抄成 0/1,再按 8/16 位分组转 ASCII(本机试过 24–58s 无有效文本,可换 0–24s / 58–74s 试试)。

待填项

答案待人工听音确认后写入 spoiler。提交时去掉空格/大小写差异,以 WeChall 服务器返回的验证结果为准。

answer (待人工听音确认后填入。)