Denoise

Overview

Clean audio by removing background noise.

Supports two methods:

'afftdn': FFT-based denoising with adjustable noise reduction (dB) 'arnndn': RNN-based denoising (requires model file in container)

Endpoint

POST https://eranol.com/api/v1/ffmpeg/audio/denoise

Request Body

{
  "url": "https://cdn.example.com/noisy-audio.mp4",
  "method": "afftdn",
  "noise_reduction": 15
}

Response Example

{
  "job_id": "f595b9ad-1e15-4452-b7c2-8f8ff09c0a34",
  "status": "pending",
  "message": "Job created. Poll status_url for progress.",
  "job_type": "caption",
  "result_url": "https://eranol.com/api/v1/ffmpeg/result/f595b9ad-1e15-4452-b7c2-8f8ff09c0a34",
  "status_url": "https://eranol.com/api/v1/ffmpeg/status/f595b9ad-1e15-4452-b7c2-8f8ff09c0a34"
}

Use Cases

1. FFT denoising — light background hum

Use afftdn with a moderate noise_reduction value to remove a consistent background hum (e.g. HVAC or fan noise) from a podcast recording.

Request Body

{
  "url": "https://cdn.example.com/podcast-raw.mp3",
  "method": "afftdn",
  "noise_reduction": 12
}

Response

{
  "job_id": "1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status": "pending",
  "message": "Job created. Poll status_url for progress.",
  "result_url": "https://eranol.com/api/v1/ffmpeg/result/1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status_url": "https://eranol.com/api/v1/ffmpeg/status/1406c589-a1f4-45d5-8745-aae07bc4de38"
}

2. FFT denoising — heavy wind noise

Apply aggressive FFT denoising to outdoor recordings with strong wind or crowd noise.

Request Body

{
  "url": "https://cdn.example.com/outdoor-interview.mp4",
  "method": "afftdn",
  "noise_reduction": 25
}

Response

{
  "job_id": "1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status": "pending",
  "message": "Job created. Poll status_url for progress.",
  "result_url": "https://eranol.com/api/v1/ffmpeg/result/1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status_url": "https://eranol.com/api/v1/ffmpeg/status/1406c589-a1f4-45d5-8745-aae07bc4de38"
}

3. RNN denoising — voice clarity

Use arnndn (neural network model) for highest-quality voice denoising on a call recording where the speaker's voice must remain natural.

Request Body

{
  "url": "https://cdn.example.com/zoom-call.mp4",
  "method": "arnndn"
}

Response

{
  "job_id": "1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status": "pending",
  "message": "Job created. Poll status_url for progress.",
  "result_url": "https://eranol.com/api/v1/ffmpeg/result/1406c589-a1f4-45d5-8745-aae07bc4de38",
  "status_url": "https://eranol.com/api/v1/ffmpeg/status/1406c589-a1f4-45d5-8745-aae07bc4de38"
}

Error Codes

Code	Meaning
401	Missing or invalid API key
402	Insufficient credits
422	Validation error — check your request body
429	Rate limit exceeded — slow down requests
500	Internal server error — try again or contact support