Everything you need to know

You can use PromptFlowStudio without an account — just drag and drop an audio file and process it. Creating a free account unlocks your personal library, lets you save results, and enables credit-based batch processing.

The Workspace is your main processing panel. It shows your current upload, processing options, and a real-time status feed.

Uploading audio

• Drag and drop any supported audio file onto the upload zone.
• Files up to 100 MB are accepted.
• Mono or stereo tracks both work; stereo is recommended for best results.
• Lossless formats (WAV, FLAC) give better separation quality than MP3.

Processing modes

Processing status

After you click Process, the workspace shows a live status:

• Queued — your job is waiting for a processing slot.
• Processing — AI model is running on your audio (30 – 120 s).
• Done — results are ready; click to listen or download.
• Failed — something went wrong; no credits are charged on failure.

Stem separation is powered by Demucs (htdemucs_ft), Facebook Research's state-of-the-art source separation model. It runs on dedicated GPU nodes to ensure consistent quality.

Output stems

Tips for best results

• Use WAV or FLAC source files — MP3 artifacts reduce separation quality.
• Stereo tracks produce better results than mono.
• Higher bitrate MP3s (256 kbps+) separate noticeably better than 128 kbps.
• Very short clips (under 10 s) may produce unstable results at stem edges.
• Songs with dense production may have bleed between stems — this is a fundamental limitation of single-model separation.

Your Library stores all processed results so you can come back, listen, and download them later. Navigate to Library in the Studio sidebar to see all your projects.

Library features

• Stream each stem directly in the browser with the built-in waveform player.
• Download individual stems or all stems as a ZIP.
• See processing status, cost in credits, and creation date for every project.
• Delete projects you no longer need to keep your library clean.

Auto-deletion policy

PromptFlowStudio uses a credit system — you buy credits upfront and spend them only when you process audio. No subscriptions, no monthly fees, credits never expire.

Credit costs

Credit packages

Top-up & Auto-refill

Go to Billing in the Studio sidebar to buy more credits. You can also enable Auto-refill — when your balance drops below a threshold, credits are automatically added using your saved payment method so you never get stuck mid-batch.

PromptFlowStudio exposes a REST API so you can integrate audio mastering into your own workflows, automate batch processing, and build custom pipelines.

Authentication

Generate an API key from Settings → API Keys in the Studio. Pass it in the Authorization header:

Authorization: Bearer pfs_your_api_key_here

Submit a job

POST /api/jobs
Content-Type: application/json
Authorization: Bearer pfs_your_api_key_here

{
  "audio_url": "https://your-storage.com/track.wav",
  "mode": "4stems"   // "4stems" | "vocals" | "karaoke" | "denoise"
}

Response:

{
  "job_id": "job_abc123",
  "status": "queued",
  "credits_charged": 10
}

Poll for results

GET /api/jobs/{job_id}
Authorization: Bearer pfs_your_api_key_here

{
  "job_id": "job_abc123",
  "status": "done",
  "result_urls": {
    "vocals": "https://cdn.../vocals.wav",
    "drums":  "https://cdn.../drums.wav",
    "bass":   "https://cdn.../bass.wav",
    "other":  "https://cdn.../other.wav"
  },
  "created_at": "2026-05-21T02:00:00Z"
}

Python example

import time, requests

API_KEY = "pfs_your_api_key_here"
BASE    = "https://promptflowstudio.click/api"

def separate_stems(audio_url: str) -> dict:
    headers = {"Authorization": f"Bearer {API_KEY}"}

    # Submit job
    r = requests.post(f"{BASE}/jobs",
        json={"audio_url": audio_url, "mode": "4stems"},
        headers=headers)
    job_id = r.json()["job_id"]

    # Poll until done
    while True:
        r = requests.get(f"{BASE}/jobs/{job_id}", headers=headers)
        data = r.json()
        if data["status"] == "done":
            return data["result_urls"]
        if data["status"] == "failed":
            raise RuntimeError("Job failed")
        time.sleep(5)

urls = separate_stems("https://your-storage.com/track.wav")
print(urls["vocals"])  # download URL for vocals stem

Access Settings from the Studio sidebar to manage your account and preferences.