[go: up one dir, main page]
CAVN AI · Open API
Agent-ready music API is live

Turn musical ideas
into songs with API
ready for your app

Use /v1/cavn/generate/prompt to transform a natural-language inspiration prompt into a complete song. CAVN AI gives apps, workflows, and AI agents a stable, composable, developer-friendly music generation layer.

API Playground
POSTPrompt-to-song generation
/v1/cavn/generate/prompt
PromptAn uplifting pop song that feels like opening the windows to a bright new morning
Modelcavn-v5.5
{
"prompt": "An uplifting pop song that feels like opening the windows to a bright new morning",
"model": "cavn-v5.5"
}
Ready742ms
The Day Unfolds.mp3Pop · uplifting · bright morning0:00 / 0:00
99.95%Service uptimeLast 30 days
< 800msAverage responseLast 30 days
Global nodesLow-latency coverageGlobal deployment
Secure by designEnterprise securityEncrypted data
Music capability layer for agents

Let agents create, process, and deliver music

Package generation, polling, stem split, MIDI conversion, and style analysis as tools agents can call. Built for MCP, OpenClaw, WorkBuddy, Hermes-style memory agents, and enterprise automation workflows.

View agent setup

Short-video BGM agent

Read a video brief, generate several BGM options, and return candidates by mood, tempo, and duration.

Game asset workflow

Batch-generate music assets by level, scene, and loop requirements while preserving task and usage records.

Brand jingle assistant

Generate multiple melody ideas from brand tone for ads, podcasts, and campaign tests.

Hermes memory loop

Persist project preferences, brand sound, and version history so agents can reuse context across sessions.

Example: generate songs from inspiration prompts

The primary path now matches the real API: POST /v1/cavn/generate/prompt. Send a natural-language prompt and model version, then use the returned song IDs to fetch results.

View docs
01Write the promptDescribe style, scene, mood, or theme.
02Create a taskReceive a task_id plus songs[].id values for async integration.
03Fetch the songUse songs[].id values to query audio and metadata.
Minimal request
curl -X POST https://api.cavn.ai/v1/cavn/generate/prompt \
  -H "Authorization: Bearer sk-live-..." \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "An uplifting pop song that feels like opening the windows to a bright new morning",
    "model": "cavn-v5.5"
  }'

API endpoints

Full docs
POST/v1/cavn/generate/promptGenerate from prompt
POST/v1/cavn/generate/customGenerate from lyrics and style
POST/v1/mureka/generate/promptMureka prompt generation
POST/v1/minimax/generate/customMiniMax music generation
POST/v1/ace-step/generate/promptACE Step prompt generation
GET/v1/music/generations/query?songid=...Fetch generation result
More endpoints are available in the full API docs

Agent tools

Translate HTTP APIs into actions that agent planners can reason about.

generate_song_from_promptCreate a song generation task from a natural-language brief.
query_generation_resultQuery async tasks and return audio URLs plus metadata.
split_audio_stemsSplit audio into vocals, drums, bass, accompaniment, and other stems.
convert_audio_to_midiConvert audio to MIDI for arrangement or sheet-music workflows.
View MCP and Skill configs

Quickstart

Generate your first piece of music in four API steps.

1

Create an account

Get your API key

2

Write the idea

Describe mood and theme

3

Send request

Call the prompt endpoint

4

Fetch result

Query by song ID and download audio

System status

All systems operational

Updated 2 minutes ago

Your API key

•••••••••••••••• a1b2

Manage keys