EU-hosted
Servers and GPU located in the EU
Remove breaths from audio & video. With one API call.
AI breath removal for podcasts, voiceovers, audiobooks, and video — at scale. Pick full, legacy, or natural mode, or mute breaths to keep the original timing. Python & JS SDKs, REST, and webhooks. EU-hosted.
- Python · JS · REST
- audio + video in/out
- EU-hosted · GDPR
live demo · one take, one flag
Hear the difference.
breath="legacy"soon
Untouched — breaths fully audible.
0:00 / 0:00tap a mark to hear a breath
> client.process(url)
One raw recording, processed with each flag. Switch mid-playback — the position is kept.
input.config.breath
Cut it. Mute it. Or let it breathe.
Three removal modes and a mute switch, set per request. Watch what each one does to the same clip.
muted=TrueSilence breaths instead of cutting them — the file keeps its exact original timing.
episode.mp34:32 → 4:18
cutcutcut
0:001:082:163:24out 4:18
speechbreath// 3 breaths — 3 cut
process.py
from cleanvoice import Cleanvoice client = Cleanvoice.from_env() result = client.process( "https://example.com/episode.mp3", breath=True,) print("Cleaned audio:", result.audio.url)Read the spec sheet.
Everything you usually scroll the docs for, before deciding if an API is worth a key.
- Endpoint
- POST api.cleanvoice.ai/v2/editsPoll GET /v2/edits/{id} or get a webhook when the file is done.
- Auth
- X-API-Key headerCreate keys at app.cleanvoice.ai — no OAuth dance.
- Breath control
- breath: true | "legacy" | "natural"Add muted: true to silence breaths instead of cutting them.
- Input formats
- Audio & videoWAV, MP3, FLAC, M4A, AAC, OGG — plus MP4, MOV, and WebM video.
- Processing speed
- ~30 s per 2–3 min clipA one-hour episode typically finishes in 5–10 minutes.
- SDKs & integrations
- Python · JavaScript · RESTPlus Make and n8n for no-code pipelines. Hosted in the EU (GDPR).
How it works
01
Get your API key
Create an account and copy your API key from the Cleanvoice dashboard.
02
Install the SDK
Use the official Python or JavaScript SDK. Alternatively, call the REST API directly.
03
Submit your file
Use a local file or url of your audio/video we should process.
04
Poll or Webhook
Jobs run async. SDKs handle polling automatically, alternatively use our webhook to upload the result to your s3 storage.
Example
Format
Why developers choose us
Ship audio features faster with APIs built for production workloads.
ISO 27001 & GDPR Compliant
DPA, SLA and on request even on premise deployment
No training on your data.
We record our own dataset in professional studios.
Simple, transparent pricing
Pay as you go or subscribe. One rate across every capability. Full pricing →
Pay as You Go
- 5 hours
$2.20/h
$11 total
- 10 hours
$2.00/h
$20 total
- 30 hours
$1.50/h
$45 total
- 200 hours
$1.00/h
$200 total
Credits valid for 2 years.
Subscription
- 10 hours / mo
$1.10/h
$11/mo
- 30 hours / mo
$1.00/h
$30/mo
- 100 hours / mo
$0.90/h
$90/mo
- 200 hours / mo
$0.88/h
$175/mo
Unused hours roll over up to 3× your plan.
200+ hours
Custom
From$0.20/ hour
Tailored feature mix and volume pricing for production workloads.
- Custom API endpoints
- Priority support
- Custom billing
Uploads are rounded up to the nearest minute.
Free credits for baby startups
Early-stage team building with audio? We’ll give you up to 200 hours of Cleanvoice API credits.
200 hours
free API credits
Use them on anything Cleanvoice does: transcription, audio cleaning, enhancement, and more. Email us a short note about your startup and what you're building, and we'll set you up personally.
Apply by emailInclude your startup name, what you’re building, and how you’d use Cleanvoice. We usually reply within 1–2 business days.
Frequently askedQuestions.
Ask NowWhat is the difference between the breath modes?
breath: true is full removal — the best default, built for general and challenging recordings. "legacy" is conservative removal that plays it safe on already-clean studio takes. "natural" lightens breaths instead of deleting every one, so the speech keeps its human rhythm. You can switch modes per request, so different shows or clients can get different treatments from the same pipeline.
Can I remove breaths without changing the timing?
Yes. Set muted: true and Cleanvoice silences the breaths instead of cutting them out. The file keeps its exact original length, which matters when the audio has to stay in sync with video, slides, or a separate music bed.
Does it work on video files?
Yes. Send MP4, MOV, or WebM and you get the edited video back — no need to extract and re-attach the audio track yourself.
Can I clean up more than breaths in the same request?
Yes — breath removal is one flag in the edit config. In the same API call you can also enable fillers, mouth_sounds, long_silences, stutters, remove_noise, studio_sound, and normalize. One upload, one result file.
How is this different from a DeBreath plugin?
Plugins like Waves DeBreath or iZotope RX run inside a DAW and need someone to open each file, tune thresholds, and listen through. The API is for pipelines: your code submits files, Cleanvoice detects and removes the breaths automatically, and a webhook tells you when each one is done. It scales from one file to thousands without an engineer in the loop.
How do I try it — is there a free tier?
Sign-up comes with free credits, so you can run real files before paying anything. After that it is pay-as-you-go or a subscription, billed by audio length. Early-stage startups can apply for up to 200 hours of free API credits.