上传音频并翻唱
- 上传一段音频,用新风格重新演绎,一步完成,异步任务
- 与「风格翻唱」不同:这里直接收
audio_url,不用 task_id - 适合用户上传自己哼唱的场景
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
固定为 suno,默认值也是 suno。
公网可直链访问的 HTTP(S) 音频地址。
本地路径、blob: 地址、以及需要登录才能打开的页面链接都不支持。建议音频长度不超过 8 分钟。
可选 v6(默认)、v6-wild、v6-mini。使用 custom_model_id 时不要填写。
创建模型任务时返回的完整 UUID。与 version、persona_id 不能同时使用。
建议明确传入。
为 false 时使用 gpt_description;为 true 时使用歌词和风格字段。
是否生成纯音乐,默认 false。
当 custom=true 且本字段为 false 时,必须填写 prompt。
灵感描述。custom=false 时必填,上限 3000 字符。
自定义模式歌词,上限 5000 字符。
风格描述,上限 1000 字符。
标题,上限 80 字符。
不希望出现的风格。
风格权重,0 到 1。
创意度权重,0 到 1。
音频权重,0 到 1。
是否对输入的歌词再做一次改编。
Male 或 Female。
Persona ID。与 custom_model_id 不能同时使用。
目标时长,10 到 360 秒。只在自定义模式下有效。
风格变化程度:off、normal、high、extra 或 max。
Max 模式,默认 false。要求使用自定义模式,费用按普通售价的 2 倍计算。
输出格式:mp3、m4a 或 wav。
Fixed at suno; the default is also suno.
An HTTP(S) audio URL that's publicly reachable as a direct link.
Local paths, blob: addresses, and page links that require login are all unsupported. Audio is best kept under 8 minutes.
v6 (default), v6-wild, or v6-mini. Don't set this when using custom_model_id.
The full UUID returned by a model creation task. Can't be combined with version or persona_id.
Best passed explicitly.
false uses gpt_description; true uses the lyrics and style fields.
Whether to generate instrumental only; defaults to false.
When custom=true and this field is false, prompt is required.
The inspiration description. Required when custom=false; up to 3000 characters.
Custom-mode lyrics; up to 5000 characters.
Style description; up to 1000 characters.
The title; up to 80 characters.
Styles you don't want.
Style weight, 0 to 1.
Creativity weight, 0 to 1.
Audio weight, 0 to 1.
Whether to rework the supplied lyrics once more.
Male or Female.
The Persona ID. Can't be combined with custom_model_id.
Target duration, 10 to 360 seconds. Effective only in custom mode.
Degree of style variation: off, normal, high, extra, or max.
Max mode; defaults to false. Requires custom mode; billed at 2× the regular price.
Output format: mp3, m4a, or wav.
与其他上传接口的分工
| 接口 | 输入 | 产出 |
|---|---|---|
| 上传音频 | audioFilePath | 只有 task_id,供后续引用 |
| 上传并翻唱(本页) | audio_url | 直接出翻唱成品 |
| 上传并延伸 | audio_url | 直接出延伸成品 |
本接口是「上传 + 翻唱」一步到位,省掉先传后引用的两步操作。
custom 建议明确传。不传时后端会按内容推断,推断结果可能与你预期不同——尤其是同时传了 prompt 和 gpt_description 的时候。
| Endpoint | Input | Output |
|---|---|---|
| Upload audio | audioFilePath | Only a task_id, for later reference |
| Upload and cover (this page) | audio_url | The finished cover, directly |
| Upload and extend | audio_url | The finished extension, directly |
This endpoint does "upload + cover" in one step, saving you the upload-then-reference two-step dance.
Pass custom explicitly. When omitted the backend infers it from the content, and the inference may not match what you expect — especially when both prompt and gpt_description are supplied.
响应
状态码。
提交状态,例如 submitted。
任务 ID,用于轮询查询结果。
成功时返回 code: 200。读取 data[0].task_id 后轮询 GET /v1/music/tasks/{task_id} 即可。
The status code.
The submission status, such as submitted.
The task ID, used to poll for the result.
On success, code: 200 is returned. Read data[0].task_id, then poll GET /v1/music/tasks/{task_id}.
音频准备
audio_url 的要求比一般素材更严:
- 必须公网直接可下载,不能要登录态
- 不能是
blob:或本地路径 - 建议不超过 8 分钟
8 分钟以上会被拒或处理质量下降。长音频先自己切出最有代表性的片段。
audio_url has stricter requirements than ordinary assets:
- Must be publicly downloadable, with no login required
- Must not be a
blob:or local path - Best kept under 8 minutes
Anything over 8 minutes is rejected or processed at reduced quality. For long audio, cut out the most representative section yourself first.
错误码
For error response formats, status codes, and retry guidance see Error codes and rate limits; for troubleshooting steps on common errors see the FAQ.
使用示例
用上传音频做翻唱
给一段公网可下载的音频,模型按提示词重新编曲演唱。audio_url 只接受 HTTP(S),本地路径和 blob: 都不行。
灵感模式重写
custom 保持默认 false 时,prompt 按创意描述理解。想要精确控制歌词时把 custom 设为 true,改用 prompt 传歌词本体。
自定义模式并具名
custom 设为 true 后,title、tags、negative_tags 才会生效。灵感模式下这几个字段会被直接丢弃。
生成纯音乐
instrumental 设为 true 输出无人声版本,适合做背景乐或伴奏。
Cover from an uploaded audio file
Give it a publicly downloadable audio URL and the model re-arranges and sings it to the prompt. audio_url accepts HTTP(S) only; local paths and blob: don't work.
Inspiration-mode rewrite
With custom left at its default false, prompt is read as a creative description. For precise lyric control set custom to true and use prompt to pass the lyrics themselves.
Custom mode with a title
Only after custom is set to true do title, tags, and negative_tags take effect. In inspiration mode these fields are discarded outright.
Generating instrumental
Setting instrumental to true outputs a vocal-free version, suited to background music or backing tracks.