Cover 改编
- 用新的风格重新演绎已有音乐，异步任务
- 用
clip_id引用源音频 - 产出的是新的 clip_id，后续操作要用新值
认证
All endpoints authenticate with a Bearer token.
Get an API key: open the console and create one.
Add it to the request header:
Authorization: Bearer YOUR_OAIRoute_KEY请求参数
模型名称固定为 flowmusic，不区分大小写。
源音频的唯一标识，取自此前成功任务的 result.music[].clip_id。
外部音频需要先上传，以换取 clip_id。
Cover 改编指令，例如「改成爵士三重奏的风格」。
这是本接口的核心，写清楚要变成什么样。只写「改成别的风格」模型会自己猜，结果不可控。
编辑强度，取值范围 0 到 1。值越大改动越大，超出范围会报错。
改编后音乐的标题。
随机种子，用于复现结果。
The model name is always flowmusic, case-insensitive.
The source audio's unique identifier, taken from result.music[].clip_id of a previous successful task.
External audio must be uploaded first to obtain a clip_id.
The cover remix instruction, e.g. "turn it into a jazz trio".
This is the core of the endpoint: state clearly what it should become. Writing only "change the style" makes the model guess, with unpredictable results.
Edit strength, 0 to 1. Higher means bigger changes; values out of range error out.
The title of the remixed music.
A random seed for reproducing results.
strength 怎么调
这是本接口最需要试的参数：
| 取值 | 效果 |
|---|---|
| 0 到 0.2 | 几乎听不出变化，只做了轻微调整 |
| 0.3 到 0.5 | 编曲变了，旋律骨架还在 |
| 0.6 到 0.8 | 明显是另一首歌，但能听出原曲影子 |
| 0.9 到 1 | 基本只剩歌词或结构，旋律几乎全新 |
建议从 0.5 起步，觉得不够就往上调，觉得面目全非就往下调。
strength 是必填项，没有默认值。不传会直接报错，不会默默取中间值。
This is the parameter that most needs experimenting:
| Value | Effect |
|---|---|
| 0 to 0.2 | Barely audible change, only slight adjustments |
| 0.3 to 0.5 | Arrangement changes, melodic skeleton remains |
| 0.6 to 0.8 | Clearly a different song, but the original still shows through |
| 0.9 to 1 | Mostly just lyrics or structure remain; the melody is almost entirely new |
Start at 0.5: raise it if the change isn't enough, lower it if the result is unrecognizable.
strength is required and has no default. Omitting it errors out rather than silently taking a middle value.
提交响应
状态码，成功时为 200。
任务状态，提交时初始为 submitted。
任务唯一标识，用于查询状态与结果。
The status code; 200 on success.
The task status; initially submitted at submission.
The unique task identifier, used to query status and results.
任务查询响应
任务 ID。
任务状态，例如 completed。
进度百分比。
创建时间戳。
完成时间戳。
实际耗时。
费用。
积分消耗。
生成的音乐列表。
music[] 元素字段
改编产物对应的新 clip_id。后续操作需要改用这个新值，不是原来的。
音乐标题。
时长，字符串形式的数值。
创建时间，ISO 8601 格式。
歌词文本。
歌词标识。
歌词时间标记，二元组数组。
音频文件地址，m4a 格式。
WAV 格式地址。
封面图地址。
The task ID.
The task status, e.g. completed.
Progress percentage.
Creation timestamp.
Completion timestamp.
Actual duration.
Cost.
Credits consumed.
The list of generated music.
Fields of music[] entries
The new clip_id of the remixed output. Later operations must use this new value, not the original.
The music title.
Duration as a numeric string.
Creation time in ISO 8601 format.
The lyrics text.
The lyrics identifier.
The lyrics timing markers, an array of pairs.
The audio file URL, in m4a format.
The WAV-format URL.
The cover image URL.
错误
错误描述，例如参数越界或余额不足。
错误类型，例如 invalid_request。
出错参数，未指明时为空串。
错误码，例如 invalid_request、quota_not_enough、rate_limit_error。
The error description, e.g. an out-of-range parameter or insufficient balance.
The error type, e.g. invalid_request.
The offending parameter, an empty string when unspecified.
The error code, e.g. invalid_request, quota_not_enough, rate_limit_error.
For more status codes and retry guidance see Error codes and rate limits; for troubleshooting steps on common errors see the FAQ.
使用示例
提交一个任务并拿到 task_id：
提交成功只表示任务已受理。真正的产物要拿返回的 task_id 去任务查询接口轮询，status 变成 completed 之后才有结果。
Submit a task and get a task_id:
A successful submission only means the task was accepted. To get the actual output, poll the task query endpoint with the returned task_id; results appear only after status becomes completed.