Pianist VS Bot

The Acoustic Turing League

[Draft] The Acoustic Turing League (ATL) Official Participant Network & Physical Temporal Auditing Interface Specification // V0.1

(草案 声学图灵联赛官方参赛方网络与物理时序审计对接规范 // 版本0.1)

1. Overview

(一、 概述)

[ENG] To ensure absolute neutrality and fairness of the inaugural “Acoustic Turing Test” league, and to preemptively eliminate cheating vectors such as pre-recorded data spoofing or covert human teleoperation via high-latency links, this tournament implements a fully asynchronous forensic pipeline with Active Jitter Injection. All participants—whether embodied AI robotics teams or human pianists—must strictly adhere to this specification.

[中文] 为确保首届“声乐图灵测试”联赛的绝对公正性,防止参赛方使用录播假数据、或通过低速链路进行幕后人类遥操作(Teleoperation)作弊,本赛事采用全链路异步取证与主动网络扰动(Active Jitter Injection)审计架构。所有参赛方(无论是具身智能机器人团队还是人类钢琴家团队)必须严格执行本规范。


2. Live Session: Active Probing & Network Alignment

(二、 赛中:反向质询与网络时序对齐)

  • [ENG] 2.1 Active Jitter Heartbeat
    • During the live stream session, the participant’s streaming host system or telemetry control node must maintain a persistent dynamic synchronization link with the Organizer’s auditing node.
      • Endpoint URL: https://api.pianist.bot/api/heartbeat?team=<your_team_id>
      • Frequency: Initiate a standard HTTP GET request every 5000 milliseconds (5s).
  • [中文] 2.1 质询心跳机制
    • 在比赛直播进行期间,参赛方托管直播页面的前端系统或现场控制节点,必须维持一条与主办方审计服务器的动态双向同步链路。
      • 端点 URL (Endpoint): https://api.pianist.bot/api/heartbeat?team=<你的战队ID>
      • 请求频次 (Frequency): 每 5000 毫秒 (5s) 发起一次标准的 HTTP GET 请求。
  • [ENG] 2.2 Jitter Injection Mechanism
    • The Organizer’s auditing node will randomly inject 50ms to 250ms of network-layer latency/jitter into the response payload (200 OK).
      The participant’s system must wait for the 200 OK response of the previous request before timing the next 5-second interval. Blind hardware timers ignoring the response loop are strictly prohibited.
  • [中文] 2.2 扰动机制说明
    • 主办方审计节点将在响应回执(200 OK)中,随机注入 50ms – 250ms 的网络物理延迟扰动。
      参赛方必须在收到上一次请求的 200 OK 响应后,方可计时并于 5 秒后发起下一次心跳。禁止采用无视回执的盲目定时器(Blind Interval)。
  • [ENG] 2.3 Recommended Connection Script in Python
    • Participant’s on-site technical team can deploy the following standard, lightweight Python script to sustain the telemetrical link. The script logs every audited timestamp and jitter value for post-match cross-examination.
  • [中文] 2.3 官方推荐:心跳挂载 Python 脚本实例
    • 参赛方现场技术人员可使用以下标准的轻量级 Python 脚本维持链路。该脚本会自动记录每一次被质询的时间戳与扰动值,用于赛后对账。

3. Post-Match: Physical Forensic Data Upload

(三、 赛后:物理存证数据流上传)

[ENG] To verify the physical authenticity of the performance, participants must stream and upload the comprehensive raw telemetric control dataset to the Organizer’s black-box archive within 180 seconds (3 minutes) after the conclusion of each musical piece.

[中文] 为了验证弹奏的物理真实性,参赛方必须在每场曲目演奏结束后的 180 秒(3 分钟)内,将原始控制数据包整体流式上传至主办方黑箱归档库。

[ENG] 3.1 API Parameters

  • Endpoint URL: https://api.pianist.bot/api/upload
  • Method: POST
  • Required Headers:
    • Content-Type: application/json or application/octet-stream
    • X-Team-ID: Participant unique identifier (e.g., TEAM_BRAND_NAME)
    • X-Auth-Token: Secure key distributed beforehand (e.g., YOUR_ENV_TOKEN)

[中文] 3.1 接口参数

  • 端点 URL (Endpoint): https://api.pianist.bot/api/upload
  • 请求方法 (Method): POST
  • 必备请求头 (Required Headers):
    • Content-Type: application/json 或 application/octet-stream
    • X-Team-ID: 参赛方唯一识别码 (e.g., 大厂战队名称)
    • X-Auth-Token: 主办方提前分发的安全密钥 (e.g., 您的环境变量密钥)

[ENG] 3.2 Payload Specification

The uploaded physical telemetry log must incorporate the following dual-stream time-series data datasets with microsecond precision:

  • Torque & Stiffness Time-Series Log: The raw physics-based force control vector text from the robotic hand (or high-fidelity capacitive keybed sensor data for human pianists).
  • Raw Acoustic Audio Sync: The precise sample-rate hash of the lossless uncompressed waveform captured by laboratory-grade near-field microphones.

[中文] 3.2 上传数据载荷规范

上传的物理日志(Log)必须包含以下两类高频时序数据,且时间戳精度必须达到 微秒级(Microsecond Precision):

  • 力矩与刚度时序流(Torque & Stiffness Log):机器手的物理力控原始文本数据(人类选手则提供现场高清传感器键床记录)。
  • 近场声学音频数据(Raw Audio Sync):现场测量级麦克风录制的无损原始音频波形采样率哈希。

4. Cross-Examination & Adjudication Rules

(四、 交叉质询与判罚标准)

[ENG] The tournament committee’s automated auditing daemon will execute a “Network-to-Physical” temporal reverse-cross-examination. Adjudication adheres strictly to the following logic:

[中文] 赛事组委会的自动化审计脚本将在赛后执行 “网络-物理”时序逆向交叉质询。判罚标准遵循以下逻辑:

  • Authentic AI / True Human: The evolutionary envelope of the physical torque time-series remains completely decoupled and statistically uncorrelated with the random Jitter anomalies injected via GET /api/heartbeat. This certifies that the performance is purely driven by local, native inference loops or localized human muscle memory.
  • Teleoperation Cheating / Remote Intervention: If the microscopic time-delay variation of the physical torque curves or specific musical note onset markers exhibits a high statistical correlation with the web-layer Jitter latency randomly applied by the Organizer, it proves a remote human operator or cloud-based intervention loop was running over the link. The committee will immediately disqualify the match and publicly release the empirical auditing matrix.
  • 真算法/真人类 (Authentic AI / True Human): 物理力矩时序波形的演进曲线,与 GET /api/heartbeat 中随机注入的 Jitter(网络抖动)时间节点完全解耦、互不相关。这证明弹奏是由本地原生算法推理或现场人类肌肉记忆直接驱动。
  • 远程遥操作作弊 (Teleoperation Cheating/Infra Interventions): 若物理力矩波形或某些特定音符的微小微调(Micro-adjustments)时序,与主办方注入的随机网络延迟呈现出 高度的统计学相关性(Statistical Correlation),则表明大厂在幕后通过网络链路进行了人类肉眼长延迟微操。组委会将直接判定该场比赛结果无效,并公开审计图表。