Base url

Welcome to the Cleanvoice API 👋

With this API you can automatically submit files to be edited by Cleanvoice, access existing edits and export them to various formats.

This is a REST API that uses predictable resource names and verbs. Our API is versioned where breaking changes are released as incremental versions. We'll try our best not to release versions too often, and to reasonably support legacy versions.

Click here if you want access to the swagger docs

If you have any questions, please write to us at We'd love to hear from you!

— Cleanvoice team


To authorize, use this code:

# With shell, you can just pass the correct header with each request
curl "" \
  -H "X-API-Key: **your_api_key_here**"

Make sure to replace **your_api_key_here** with your API key.

Cleanvoice uses API keys to allow access to the API. You can get your API key by logging in and going to settings

Cleanvoice expects the API key to be included in all API requests to the server in a header that looks like the following:

X-Api-Key: **your_api_key_here**


To get started, you have to provide an media file. Check here for the currently supported audio files.

There are two ways to upload your audio files(s): a media file URL (accesible to our servers) or direct file upload via the API. If you have a media file URL, skip this chapter.

Upload via file URL

If you choose this option, a media file URL must be provided. This can be a publicly accessible URL hosted by yourselves or a third-party. We also support public links from services like YouTube and Vimeo.Your files are directly stored in our (DigitalOcean S3 bucket). The API will return a link, which you will need when starting a new edit.

To upload files to our DO S3 bucket, you must first request a signed URL using the endpoints below.

1. Get a Signed Url

curl -X 'POST' \
  '' \
  -H 'X-API-Key: X-API-Key: **your_api_key_here**' 

The above command returns JSON structured like this:

  "signedUrl": "\?X-Amz-Algorithm..."

This endpoint returns a signed URL which can be used to make PUT requests to our S3 Digitalocean bucket. More information here:

Once the file is uploaded, this same url should be used as the tmp_url when creating the associated edit.

HTTP Request



Parameter Description
filename (required) The filename and extension of the media file (e.g. my_media.mp3)

2. Upload your file with the Signed Url

curl -X PUT -T my_media.mp3 -L "https://signed-url..."

Simple upload to an DO S3 Bucket

Once you have a signed url, you have to upload the file there, and you do it as you would with any S3 bucket.

To upload a file to the signed URL you may use cURL like the example on the right.

Here you have more information about how to deal with Digitalocean S3 signed urls


Create a Edit

curl -X POST "" \
  -H "X-API-Key: **your_api_key_here**" \
  -H "Content-Type: application/json" \
  -d '{
  "input": {
    "files": [
    "config": {}

The above command returns JSON structured like this:

  "id": "9a14b8fe-3604-43e4-8c06-06a7491ae8fc"

This endpoint creates a new edit. After a edit is created, the system will proceed to automatically editing it. You can check if the editing process has finished by retrieving a edit.

Cleanvoice requires a public link to edit your audio, if this is not possible go back to the upload section. If you have allready uploaded your audio via our api, you can use the signed URL instead.

Note all uploads and edits are stored for 7 days only. If you require longer storage, write us.

The API can edit multi-track audio as well. Multi-track uploads are where every audio file contains a different speaker. To use it, just pass multiple files. Single-Track are where all the speakers are merged into a single file.

HTTP Request




Parameter Type Description
files Array (required) A list of URL-Strings containing Audio. For Single-Track, a single link is located in the array. Else Cleanvoice assumes it's a multi-track audio recording


Parameter Type Description
send_mail Boolean Send an email to your account with the edited file(s). Default: False
timestamps_only Boolean Send only the JSON with the edits, without editing the audio file. (Faster Processing) Default: False

Retrieve a Edit

curl "<ID>" \
  -H "Authorization: Bearer **your_api_key_here**"

The above command returns JSON structured like this:

  "status": "SUCCESS",
  "result": {
    "download_url": "",
    "edits": {
      "edits": [
          "end": 11.86,
          "label": "DEADAIR",
          "start": 9.86
          "end": 16.38,
          "label": "FILLER_SOUND",
          "start": 15.53
          "end": 54.95,
          "label": "MOUTH_SOUND",
          "start": 53.95
          "end": 58.95,
          "label": "MOUTH_SOUND",
          "start": 56.95

    "statistics": {
      "DEADAIR": 1,
      "REPHRASE": 0,
      "WORD_DUB": 0,
      "LONG_PAUSE": 0,
      "STUTTERING": 0,
      "MOUTH_SOUND": 2,
      "FILLER_SOUND": 1,
      "NORMAL_PAUSE": 0,
    "id": "107b9579-583e-4585-bfa0-6a76cd15a4bb",
    "task_id": "9a14b8fe-3604-43e4-8c06-06a7491ae8fc",
    "filename": "jyYweGY_audio.mp3",
    "length": 5.899319449999999
  "task_id": "9a14b8fe-3604-43e4-8c06-06a7491ae8fc"

This endpoint retrieves information about a specific edit. The results can be used in two ways.

HTTP Request


Edits State Descriptions

Value Description
PENDING Waiting for execution or unknown edit ID
STARTED Edit has been started
SUCCESS Audio has been successfully edited
RETRY Task is being retried
FAILURE File failed to process


We currently support the following fileformats: For certain formats, we can't return the original codec or audio quality.

fileformat Description Returned fileformat the same?
mp3 MPEG-1 Audio Layer-3
wav Waveform Audio File Format
flac Free Lossless Audio Codec


We support the following languages:

Code Language Support
EN English Full
DE German Full
RO Romanian Full
FR French Partial
NL Dutch Partial
BG Bulgarian (Bulgaria) Partial
AR Arabic (United Arab Emirates) Partial
tr-TR Turkish (Turkey) Partial

Note that, we don't work on the word level. We work on the phonetical level, therefore if theres a language which is similar to two languages which we have listed, it could work as well. Say German and English are supported, then Dutch should work as well.