AI Agent Hub
Back to skills
Baiyin AI Singer Cover icon

Baiyin AI Singer Cover

AI Agent Updated 2026.08.29

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_7b3c719e/baiyin-cover-sing-skill into your AI assistant.

About this skill

Problem

In AI agent conversations, users often ask to “cover this song,” “check the last task,” or “give me the audio,” but a cover workflow still involves model selection, public audio handling, parameter defaults, async polling, and result validation. baiyin-cover-sing-skill turns these steps into a stable execution path, reducing the chance that the agent guesses modelId, invents taskId, or delivers garbled output as success.

How it works

  • Pick the model first: Before creating a task, the skill requires an explicit modelId. If missing, it calls GET /api/open/v1/cover/models and presents modelId, modelName, description, and demoUrl when available.
  • Fill parameters conservatively: It extracts taskName, inputFullUrl, type, reverbType, format, and pitchShift from the conversation. Optional fields fall back to defaults such as type = 0, reverbType = 3, and format = mp3, so the agent does not keep asking for non-blocking details.
  • Handle public audio: If the user provides a local file, attachment, or private link, the skill guides the flow through Baiyin's upload endpoint to obtain a public URL instead of asking the user to prepare OSS or CDN storage manually.
  • Create and poll: It calls POST /api/open/v1/cover/sing to create the task, then returns taskId and status. Queries use GET /api/open/v1/tasks/{taskId}; polling may continue on queued / processing and stop on succeeded / failed.
  • Validate output: On success, it prioritizes audioUrl, inputVocalUrl, inputInstrumentalUrl, inferenceUrl, format, and score. If the response contains ??, garbled text, or a mismatched task name, it treats the result as abnormal rather than final.

Boundaries

This skill targets AI singer cover on the Baiyin Open Platform, not singer training, general TTS, or local inference. It requires BAIYIN_API_KEY and access to https://ai.hikoon.com. When the request strongly implies a male-to-female or female-to-male voice shift, the skill reminds the user of suggested pitchShift values, such as +12 or -12, but only writes the value after confirmation or explicit user input.

Use Cases

  • Select a singer model in an agent conversation and create an AI cover task via the Baiyin API.
  • Convert local files or private audio links into public URLs through Baiyin upload before creating a task.
  • Query the latest cover task status and return the final audio plus vocal and instrumental URLs on success.
  • Remind the user of the suggested pitchShift value for male-to-female or female-to-male cover requests.

Best For

  • AI agent developers integrating the Baiyin Open Platform who need conversational cover task creation and querying.
  • Creators assembling music demos or cover material who need quick AI singer covers and downloadable audio links.
  • Engineers debugging voice or music APIs who need consistent model selection, defaults, and error-code handling.
  • Agent product managers building multi-turn task tracking who need status polling and clean success-field delivery.