1. Generate and save to a project
Send your API key and numeric user ID in the JSON body. Text and voice_name are required. Optional project_id defaults to Home; use an ID from your projects list for another folder. Speed, pitch, and a supported speaking style are optional.
POST https://crikk.com/generate-audio/v2
Content-Type: application/json
{
"api_token": "YOUR_API_TOKEN",
"user_id": 42,
"text": "Hello, world! [1s] Welcome to Crikk.",
"voice_name": "en-US-AriaNeural",
"project_id": "home"
}
// 202 → { "job_id": "RETURNED_JOB_ID", "status": "pending", ... }
Audio saves automatically when generation completes, even if you never poll. Its characters are charged once against the same API allowance used by MCP.
2. Check the job and get the audio URL
Poll with the returned job_id and both API-key headers. Completed responses contain JSON with an audio_url. Open that URL to download, or use it in your audio player.
GET https://crikk.com/generate-audio/v2/status/RETURNED_JOB_ID
Authorization: Bearer YOUR_API_TOKEN
X-Crikk-User-Id: YOUR_NUMERIC_USER_ID
// 200 while generating → { "job_id": "...", "status": "pending" }
// 200 when saved → JSON containing:
// status: "completed", id, project_id, title, text, voice,
// audio_url: "https://files.crikk.com/EXAMPLE.mp3",
// mime_type: "audio/mpeg"
Polling records last ten minutes from submission and may be lost on a server restart. Saved audio remains in your library under its normal retention rules. Repeated status checks do not charge again.
3. Browse and listen again
List projects to choose a folder. List audio without parameters for Home, or pass project_id and page. Each page returns up to 20 recordings, newest first. Increment page while has_next_page is true. Browsing uses no speech characters.
GET https://crikk.com/generate-audio/library/projects
GET https://crikk.com/generate-audio/library/audio?project_id=home&page=1
// Both requests require:
Authorization: Bearer YOUR_API_TOKEN
X-Crikk-User-Id: YOUR_NUMERIC_USER_ID
// Audio response: { project_id, files, page, has_next_page }
// Each file includes an audio_url for listening or downloading.
Shared generation settings
API and MCP use the free voice catalog and fixed MP3 output at 48 kHz, 192 kbps, mono. Each pause is capped at 10 seconds; the shared Microsoft synthesis timeout is 50 minutes. Output format, timeout, and maximum pause are managed by the server.
Existing clients can keep using the original /generate-audio binary download flow. Use /generate-audio/v2 for library saves. MCP supports the same saved library, with tools to list projects, list audio, find voices, generate speech, and retrieve results.
