Enhance Voice

POST

Studio-quality voice enhancement using DeepFilterNet (ML) — removes broadband noise underneath speech, the kind plain FFmpeg filters can't separate. Adds optional EQ voice profiles and loudness normalization. Comparable to Adobe "Enhance Speech". Works on audio or video; video in returns enhanced video out by default. Example: {"url": "https://cdn.example.com/raw.mp4", "voice_profile": "bright", "keep_video": true}

Overview

Studio-quality voice enhancement using DeepFilterNet, a trained ML model that removes broadband noise underneath speech.

After noise removal it applies an optional EQ voice profile and loudness normalization, then encodes the result. Video inputs are muxed back onto the original (copied, not re-encoded) video track and returned as MP4; audio-only inputs return audio.

This is an async job: POST to create a job, then poll the status_url until it completes.

Endpoint

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

Request Body

{
  "url": "https://cdn.example.com/raw.mp4",
  "voice_profile": "bright",
  "loudness_target": -16,
  "post_filter": true,
  "atten_limit_db": 100,
  "keep_video": true,
  "output_format": "mp3"
}

Parameters

Field	Type	Default	Description
url	string (URL)	required	Remote audio or video file to enhance.
voice_profile	string	"bright"	EQ tone applied after noise removal: none (flat), natural (broadcast tone), deep (fuller voice), bright (max clarity for phone speakers — default).
loudness_target	number | null	-16	Target loudness in LUFS. -16 = Spotify/Apple, -14 = YouTube. Range -31 to -5. Set null to skip normalization.
post_filter	boolean	true	Enable DeepFilterNet post-filter for stronger noise attenuation. Keep on for noisy sources.
atten_limit_db	number	100	Noise attenuation limit in dB. Range 0 to 100. 100 = full removal; lower blends back some original signal for a more natural result.
keep_video	boolean	true	If input is video, mux enhanced audio back onto the original video and return MP4. Set false to force audio-only output. Audio-only inputs always return audio.
output_format	string	"mp3"	Audio format when keep_video is false: mp3 (320 kbps), m4a (AAC 192 kbps), or wav (lossless).

Response Example

{
  "job_id": "9f8b2c1a-7d4e-4a2b-b1c3-0e5f6a7b8c9d",
  "status": "pending",
  "message": "Job created. Poll status_url for progress.",
  "job_type": "enhance",
  "result_url": "https://eranol.com/api/v1/ffmpeg/result/9f8b2c1a-7d4e-4a2b-b1c3-0e5f6a7b8c9d",
  "status_url": "https://eranol.com/api/v1/ffmpeg/status/9f8b2c1a-7d4e-4a2b-b1c3-0e5f6a7b8c9d"
}

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

---

Poll Status

GET

Get the current status of an enhance job. Poll this until status is completed, then read the result object.

Endpoint

GET https://eranol.com/api/v1/ffmpeg/status/{job_id}

Status Values

pending → downloading → processing → completed / failed / cancelled

Progress Stages

"Extracting audio (1/4)" → "Enhancing voice with DeepFilterNet (2/4)" → "Applying 'bright' EQ + loudness (3/4)" → "Encoding output (4/4)"

Response Example (completed)

{
  "job_id": "9f8b2c1a-7d4e-4a2b-b1c3-0e5f6a7b8c9d",
  "job_type": "enhance",
  "status": "completed",
  "progress": null,
  "created_at": 1750684800.12,
  "started_at": 1750684803.44,
  "completed_at": 1750684820.71,
  "result": {
    "url": "https://eranol.com/outputs/enhanced-9f8b2c1a.mp4",
    "filename": "enhanced-9f8b2c1a.mp4",
    "format": "mp4",
    "duration": 30.0,
    "audio_codec": "aac",
    "sample_rate": "48000",
    "resolution": "1280x720",
    "engine": "deepfilternet",
    "voice_profile": "bright",
    "loudness_target": -16.0,
    "post_filter": true,
    "keep_video": true
  }
}

Result Fields

Field	Type	Description
url	string	Public download URL of the enhanced file.
filename	string	enhanced-<uuid>.<ext>
format	string	mp4 (video kept) or mp3 / m4a / wav.
duration	number	Length in seconds.
audio_codec	string	e.g. aac, mp3, pcm_s16le.
sample_rate	string	e.g. "48000".
resolution	string | null	"WxH" for video, null for audio-only.
engine	string	Always "deepfilternet".

Error Codes

Code	Meaning
401	Missing or invalid API key
404	Job not found
500	Internal server error — try again or contact support