AI Music

The Specific Problem

Music creation often demands professional skills and time, especially in fast-paced content scenarios like short videos, podcasts, or advertisements, where creators need customized yet efficient music solutions. Traditional methods rely on manual composition or stock libraries, struggling to balance originality and speed. AI music generation tools automate this via natural language descriptions, but implementing them involves technical challenges like API authentication, CLI integration, and error recovery, requiring developers to handle underlying details for reliable use.

How the Skill Works

This skill is built on Sponge Music's AI services, providing a command-line interface (CLI) tool aimv to generate songs and BGM from descriptions. Core capabilities include:

• Generating Custom Music: Users input descriptions such as theme, style, mood, or lyrics; the CLI calls the API to produce audible audio works, supporting both songs and background music.
• Managing Works and Accounts: Generated works automatically sync to a Sponge Music account, interoperable with mini-programs or web platforms (lexuan.club) for easy editing and downloading.
• Automated Workflows: Agents can directly execute shell commands, like node /bin/aimv.js init for initialization or aimv song create for generation, enabling end-to-end operations.

Key steps involve:
1. Initialization and Authentication: On first use, run node /bin/aimv.js init to get a connection link; users log in and provide an API Key for binding (one-time setup).
2. Music Generation: Execute generation commands based on requirements; the CLI returns structured responses, and the Agent handles status to guide users through preview or download.
3. Error Handling and Recovery: The CLI outputs tagged error messages, such as [error] or [hint], and Agents must act based on codes (e.g., auth_required, insufficient_points)—for instance, sending login or payment links.

In restricted environments (e.g., sandboxes), set the AIMV_FORCE_CURL=1 environment variable to let the CLI switch to the system curl as a transport layer, ensuring network availability; in extreme cases, directly calling the API with curl is possible, following the same response envelope format (e.g., code, data fields).

Boundaries and Considerations

• Dependency on External Services: Generation fully relies on Sponge Music's API, requiring stable internet; points and membership systems affect usage quotas, necessitating top-ups via payment links when insufficient.
• Language and Scenario Limits: Primarily optimized for Chinese music generation, with limited support for English or other languages; suitable for quick creative prototyping, though professional production may still need manual adjustments.
• Security and Configuration: API Keys are long-term valid and should be safeguarded; the config directory can be redirected via AIMV_CONFIG_DIR to adapt to different environments.
• Error Recovery Mechanism: Agents must parse the CLI's structured tags (e.g., code=quota_limited) rather than simply relaying errors, providing accurate user guidance.
• Work Management: After generation, each project defaults to two candidate songs, both requiring presentation to users; reference audio upload is supported but through specific API endpoints.