分开音色参考与风格引导，避免把续说模式当同一接口
本页展示随包原生输入和验收参考。目标产品运行状态：未运行。表格与图片不代表产品已经生成的结果。
input.csv / practice-data.json
本课实际输入字段
| condition | reference | style | targetText |
|---|---|---|---|
| reference-only | original synthetic sample | ||
| reference-and-style | same synthetic sample | slightly faster, cheerful tone | |
| The workshop starts at ten; the order is not approved. |
先运行参考音色模式
reference.csv：作者验收映射，不是目标API回包
| field | expected | comparison |
|---|---|---|
| ten | time unchanged | |
| not approved | negation retained | |
| record separately with actual audio | speaker reference and style |
应当保留的约束
| field | expected | comparison |
|---|---|---|
| ten | time unchanged | |
| not approved | negation retained | |
| record separately with actual audio | speaker reference and style |
故意构造的错误样例
| mode | claim | fault | text |
|---|---|---|---|
| continuation | same interface contract | different generation mode | |
| meaning changed | order approved |
先保存真实产品输出，再逐项对照。actualOutput / actualResult 在未执行时保持 not_run，不能把这里的期望值复制成已测结果。