Audio tools

Clean audio removes background noise, splits a song into vocals and music, splits stems, or isolates a described sound.
Voice changer keeps a recording's words, timing and delivery in another voice.

Both run in the background on an uploaded audio or video file.

Endpoints:

GET /v1/clean-audio/options
POST /v1/clean-audio
GET /v1/clean-audio/{id}
GET /v1/clean-audio/{id}/files/{name}
GET /v1/voice-changer/options
POST /v1/voice-changer
GET /v1/voice-changer/{id}
GET /v1/voice-changer/{id}/files/{name}

Clean audio

Get clean audio options

GET /v1/clean-audio/options

Response

200 OK

json

{
  "modes": [
    { "id": "voice", "label": "Remove background noise", "description": "Keep only the voice: noise, music and room sound removed. Works on video too.", "creditsPerMinute": 1000 },
    { "id": "vocals", "label": "Vocals and music", "description": "Split a song into the singing and the instrumental.", "creditsPerMinute": 600 },
    { "id": "stems", "label": "Split into stems", "description": "Vocals, drums, bass and other instruments, or 6 stems with guitar and piano.", "creditsPerMinute": 400 },
    { "id": "sound", "label": "Isolate a sound", "description": "Describe any sound, like a dog barking or applause, and pull it out.", "creditsPerMinute": 600 }
  ],
  "stemSets": [
    { "id": "4", "stems": ["Vocals", "Drums", "Bass", "Other instruments"] },
    { "id": "6", "stems": ["Vocals", "Drums", "Bass", "Guitar", "Piano", "Other instruments"] }
  ],
  "soundExamples": ["a person speaking", "dog barking", "applause", "birds singing", "car horn", "acoustic guitar"],
  "maxPromptChars": 200,
  "maxSeconds": 3600,
  "maxUploadBytes": 2147483648
}

Credits

Free.

Example

bash

curl https://api.cinara.ai/v1/clean-audio/options \
  -H "Authorization: Bearer $CINARA_API_KEY"

Clean audio

POST /v1/clean-audio

Starts a clean audio job on an uploaded file.

Needs a verified phone.

Request body

Field	Type	Required	Description
`key`	string	yes	The storage key of a finished audio or video upload.
`mode`	string	yes	`voice`, `vocals`, `stems` or `sound`.
`prompt`	string	`sound` mode	The sound to isolate, 1–200 characters. Ignored in other modes.
`stems`	string	no	`stems` mode: `"4"` (default) or `"6"`. Any other value is `"4"`.
`keepVideo`	boolean	no	`voice` mode on a video: also return the video with the clean audio. Default `true`.
`durationSeconds`	number	no	A hint: the file's length in seconds, up to 3,600. Cinara reads the real length from the file and uses this only when it can't.
`fileName`	string	no	Up to 200 characters. Defaults to the upload's file name.

Steps by mode:

Mode	Audio file	Video file
`voice`	Remove background noise	Remove background noise, then put the clean audio back on the video (with `keepVideo`)
`vocals`, `sound`	Separate the sound	Take the audio out of the video, then separate
`stems`	Split into stems	Take the audio out of the video, then split

Response

201 Created

json

{
  "generation": {
    "id": "5c8a2e6f-9d1b-4f3a-b7e4-6a0d3c9f2b18",
    "module": "clean_audio",
    "status": "running",
    "input": {
      "key": "workspaces/3f6b2a1c-8d4e-4b7a-9c2d-5e1f0a9b8c7d/uploads/1e7c4a9b-5d2f-4b6e-8a3c-7f0b9d2e5a64/demo-track.wav",
      "fileName": "demo-track.wav",
      "contentType": "audio/wav",
      "size": 37044000,
      "durationSeconds": 210,
      "durationMeasured": true,
      "mode": "stems",
      "prompt": null,
      "stems": "4",
      "keepVideo": false,
      "totalSteps": 1
    },
    "output": { "stage": "queued", "step": 1, "stageLabel": "Splitting into stems" },
    "credits": 1400,
    "error": null,
    "created_at": "2026-09-15T18:02:15.550913+00:00",
    "completed_at": null
  },
  "credits": 43831
}

Errors

Status	`error`	Message
400	`invalid_request`	`Upload a file first`
400	`invalid_request`	`Choose what to do with the audio`
400	`invalid_request`	`Describe the sound in 1 to 200 characters`
400	`invalid_request`	`Files can be up to 1 hour long`
404	`not_found`	`That upload isn't finished. Upload the file again.`
400	`invalid_request`	`Upload an audio or video file`
402	`insufficient_credits`	`This needs 1,400 credits`
403	`phone_unverified`	`Verify your phone number to start creating.`
502	`generation_failed`	`Clean audio couldn't start. Your credits were refunded.`

The length is read from the file (MP3, WAV, MP4, M4A or MOV) before anything is charged, and a file over 1 hour is refused. When it can't be read (for example WebM, OGG or FLAC), a durationSeconds over 1 hour is refused instead.

Credits

max(1, ceil(minutes × rate)), where the rate per minute is:

Mode	Credits per minute
`voice`	1,000
`vocals`	600
`stems`	400
`sound`	600

Length read from the file: charged on the file's real length. input.durationMeasured is true.
Length can't be read: a reserve is held on the longer of durationSeconds and an estimate from the file size, up to 1 hour. When the job finishes, the charge is settled to the length of the result. input.durationMeasured is false.

Refunded if it fails.

Example

bash

curl -X POST https://api.cinara.ai/v1/clean-audio \
  -H "Authorization: Bearer $CINARA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"key": "workspaces/3f6b2a1c-8d4e-4b7a-9c2d-5e1f0a9b8c7d/uploads/1e7c4a9b-5d2f-4b6e-8a3c-7f0b9d2e5a64/demo-track.wav", "mode": "stems", "stems": "4", "durationSeconds": 210}'

Get a clean audio job

GET /v1/clean-audio/{id}

Checks a clean audio job, moves it forward, and returns it with its files once finished.

Response

200 OK

json

{
  "generation": {
    "id": "5c8a2e6f-9d1b-4f3a-b7e4-6a0d3c9f2b18",
    "module": "clean_audio",
    "status": "succeeded",
    "input": {
      "key": "workspaces/3f6b2a1c-8d4e-4b7a-9c2d-5e1f0a9b8c7d/uploads/1e7c4a9b-5d2f-4b6e-8a3c-7f0b9d2e5a64/demo-track.wav",
      "fileName": "demo-track.wav",
      "contentType": "audio/wav",
      "size": 37044000,
      "durationSeconds": 210,
      "durationMeasured": true,
      "mode": "stems",
      "prompt": null,
      "stems": "4",
      "keepVideo": false,
      "totalSteps": 1
    },
    "output": { "step": 1 },
    "credits": 1400,
    "error": null,
    "created_at": "2026-09-15T18:02:15.550913+00:00",
    "completed_at": "2026-09-15T18:03:40.118822+00:00"
  },
  "files": [
    { "name": "vocals", "label": "Vocals", "contentType": "audio/mpeg", "size": 3371264, "url": "https://api.cinara.ai/media/1789539820/Ee4tA…/workspaces/…/vocals.mp3" },
    { "name": "drums", "label": "Drums", "contentType": "audio/mpeg", "size": 3371264, "url": "https://api.cinara.ai/media/1789539820/Uu0kM…/workspaces/…/drums.mp3" },
    { "name": "bass", "label": "Bass", "contentType": "audio/mpeg", "size": 3371264, "url": "https://api.cinara.ai/media/1789539820/Rw9bN…/workspaces/…/bass.mp3" },
    { "name": "other", "label": "Other instruments", "contentType": "audio/mpeg", "size": 3371264, "url": "https://api.cinara.ai/media/1789539820/Lc2hS…/workspaces/…/other.mp3" }
  ],
  "mediaUrl": "https://api.cinara.ai/media/1789539820/Mf6jQ…/workspaces/…/uploads/…/demo-track.wav",
  "credits": 43831
}

Files by mode (empty until the job succeeds):

Mode	Files (`name`: label)
`voice`	`voice`: Clean voice. On a video with `keepVideo`, also `video`: Clean video.
`vocals`	`vocals`: Vocals; `music`: Music.
`sound`	`sound`: the prompt, capitalised; `rest`: Everything else.
`stems` `"4"`	`vocals`, `drums`, `bass`, `other`
`stems` `"6"`	`vocals`, `drums`, `bass`, `guitar`, `piano`, `other`

mediaUrl is a signed link (6 hours) to the uploaded file.

Failures: Clean audio took too long and was stopped., This file couldn't be read. Try MP3, WAV, M4A, MP4 or WebM., Clean audio failed.

Errors

Status	`error`	Message
404	`not_found`	`Not found`

Credits

Free to read.

Polling example

bash

while :; do
  RES=$(curl -s "https://api.cinara.ai/v1/clean-audio/$ID" -H "Authorization: Bearer $CINARA_API_KEY")
  STATUS=$(echo "$RES" | jq -r .generation.status)
  [ "$STATUS" != "running" ] && break
  echo "Step $(echo "$RES" | jq -r .generation.output.step) of $(echo "$RES" | jq -r .generation.input.totalSteps): $(echo "$RES" | jq -r .generation.output.stageLabel)"
  sleep 5
done
echo "$RES" | jq -r '.files[] | "\(.name) \(.url)"'

Download a clean audio file

GET /v1/clean-audio/{id}/files/{name}

Downloads one finished file as an attachment.

Response

200 OK with the file and Content-Disposition: attachment; filename="demo-track-vocals.mp3".

Errors

Status	`error`	Message
404	`not_found`	`File not found`

Credits

Free.

Example

bash

curl -L "https://api.cinara.ai/v1/clean-audio/5c8a2e6f-9d1b-4f3a-b7e4-6a0d3c9f2b18/files/vocals" \
  -H "Authorization: Bearer $CINARA_API_KEY" -o vocals.mp3

Voice changer

Get voice changer options

GET /v1/voice-changer/options

Response

200 OK

json

{
  "creditsPerMinute": 300,
  "noiseCreditsPerMinute": 1000,
  "maxSeconds": 600,
  "maxUploadBytes": 2147483648
}

Credits

Free.

Example

bash

curl https://api.cinara.ai/v1/voice-changer/options \
  -H "Authorization: Bearer $CINARA_API_KEY"

Change a voice

POST /v1/voice-changer

Starts changing the voice in an uploaded recording.

Needs a verified phone.

A library voice converts toward its preview clip. A workspace voice converts toward its saved sample.

Request body

Field	Type	Required	Description
`key`	string	yes	The storage key of a finished audio or video upload.
`voiceId`	string	yes	A library voice id, or a workspace voice's `providerVoiceId`.
`removeNoise`	boolean	no	Remove background noise first. Default `false`.
`keepVideo`	boolean	no	For videos: also return the video with the new voice. Default `true`.
`durationSeconds`	number	no	A hint: the recording's length, up to 600 seconds. Cinara reads the real length from the file and uses this only when it can't.
`fileName`	string	no	Up to 200 characters. Defaults to the upload's file name.

Steps: remove noise (with removeNoise) or take the audio out of a video, then change the voice, then put the new voice on the video (videos with keepVideo).

Response

201 Created

json

{
  "generation": {
    "id": "e3b7d1a9-4c6f-4e2b-9a8d-5f1c0e7b3a26",
    "module": "voice_changer",
    "status": "running",
    "input": {
      "key": "workspaces/3f6b2a1c-8d4e-4b7a-9c2d-5e1f0a9b8c7d/uploads/8a3d6f1c-2e9b-4d7a-b5c8-1f4e7a0d9c32/promo-read.mp3",
      "fileName": "promo-read.mp3",
      "contentType": "audio/mpeg",
      "size": 720000,
      "durationSeconds": 45,
      "durationMeasured": true,
      "voiceId": "captain",
      "voiceName": "Captain · Ryan",
      "targetContentType": null,
      "removeNoise": true,
      "keepVideo": false,
      "totalSteps": 2
    },
    "output": { "stage": "queued", "step": 1, "stageLabel": "Removing background noise" },
    "credits": 975,
    "error": null,
    "created_at": "2026-09-15T18:30:05.887452+00:00",
    "completed_at": null
  },
  "credits": 43006
}

Errors

Status	`error`	Message
400	`invalid_request`	`Upload a recording first`
400	`invalid_request`	`Choose a voice`
400	`invalid_request`	`Recordings can be up to 10 minutes`
404	`not_found`	`That upload isn't finished. Upload the file again.`
400	`invalid_request`	`Upload an audio or video file`
402	`insufficient_credits`	`This needs 975 credits`
403	`phone_unverified`	`Verify your phone number to start creating.`
502	`generation_failed`	`Voice change couldn't start. Your credits were refunded.`

The length is read from the file (MP3, WAV, MP4, M4A or MOV) before anything is charged, and a recording over 10 minutes is refused. When it can't be read (for example WebM, OGG or FLAC), a durationSeconds over 10 minutes is refused instead.

Credits

max(1, ceil(minutes × rate)). The rate is 300 credits per minute, or 1,300 (300 + 1,000) with removeNoise.

Length read from the file: charged on the recording's real length.
Length can't be read: a reserve is held on the longer of durationSeconds and an estimate from the file size, up to 10 minutes. When the job finishes, the charge is settled to the length of the result.

Refunded if it fails.

Example

bash

curl -X POST https://api.cinara.ai/v1/voice-changer \
  -H "Authorization: Bearer $CINARA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"key": "workspaces/3f6b2a1c-8d4e-4b7a-9c2d-5e1f0a9b8c7d/uploads/8a3d6f1c-2e9b-4d7a-b5c8-1f4e7a0d9c32/promo-read.mp3", "voiceId": "captain", "removeNoise": true, "durationSeconds": 45}'

Get a voice changer job

GET /v1/voice-changer/{id}

Checks a voice changer job, moves it forward, and returns it with its files once finished.

Response

200 OK

json

{
  "generation": {
    "id": "e3b7d1a9-4c6f-4e2b-9a8d-5f1c0e7b3a26",
    "module": "voice_changer",
    "status": "succeeded",
    "input": { "key": "workspaces/…/uploads/…/promo-read.mp3", "voiceId": "captain", "voiceName": "Captain · Ryan", "totalSteps": 2, "…": "…" },
    "output": { "step": 2 },
    "credits": 975,
    "error": null,
    "created_at": "2026-09-15T18:30:05.887452+00:00",
    "completed_at": "2026-09-15T18:30:51.340126+00:00"
  },
  "files": [
    { "name": "voice", "label": "Captain · Ryan's voice", "contentType": "audio/wav", "size": 3969080, "url": "https://api.cinara.ai/media/1789541451/Yx3pF…/workspaces/…/voice.wav" }
  ],
  "mediaUrl": "https://api.cinara.ai/media/1789541451/Gd8sK…/workspaces/…/uploads/…/promo-read.mp3",
  "credits": 43006
}

Files: voice (labelled with the voice's name), and for videos with keepVideo, video ("Video with the new voice"). Empty until the job succeeds.

Failures: Voice change took too long and was stopped., This file couldn't be read. Try MP3, WAV, M4A, MP4 or WebM., Voice change failed.

Errors

Status	`error`	Message
404	`not_found`	`Not found`

Credits

Free to read.

Polling example

bash

while :; do
  RES=$(curl -s "https://api.cinara.ai/v1/voice-changer/$ID" -H "Authorization: Bearer $CINARA_API_KEY")
  [ "$(echo "$RES" | jq -r .generation.status)" != "running" ] && break
  sleep 5
done
curl -L "https://api.cinara.ai/v1/voice-changer/$ID/files/voice" -H "Authorization: Bearer $CINARA_API_KEY" -o new-voice.wav

Download a voice changer file

GET /v1/voice-changer/{id}/files/{name}

Downloads one finished file (voice or video) as an attachment.

Response

200 OK with the file and Content-Disposition: attachment; filename="promo-read-voice.wav".

Errors

Status	`error`	Message
404	`not_found`	`File not found`

Credits

Free.

Example

bash

curl -L "https://api.cinara.ai/v1/voice-changer/e3b7d1a9-4c6f-4e2b-9a8d-5f1c0e7b3a26/files/voice" \
  -H "Authorization: Bearer $CINARA_API_KEY" -o new-voice.wav

Audio tools ​

Clean audio ​

Get clean audio options ​

Response ​

Credits ​

Example ​

Clean audio ​

Request body ​

Response ​

Errors ​

Credits ​

Example ​

Get a clean audio job ​

Response ​

Errors ​

Credits ​

Polling example ​

Download a clean audio file ​

Response ​

Errors ​

Credits ​

Example ​

Voice changer ​

Get voice changer options ​

Response ​

Credits ​

Example ​

Change a voice ​

Request body ​

Response ​

Errors ​

Credits ​

Example ​

Get a voice changer job ​

Response ​

Errors ​

Credits ​

Polling example ​

Download a voice changer file ​

Response ​

Errors ​

Credits ​

Example ​

Audio tools

Clean audio

Get clean audio options

Response

Credits

Example

Clean audio

Request body

Response

Errors

Credits

Example

Get a clean audio job

Response

Errors

Credits

Polling example

Download a clean audio file

Response

Errors

Credits

Example

Voice changer

Get voice changer options

Response

Credits

Example

Change a voice

Request body

Response

Errors

Credits

Example

Get a voice changer job

Response

Errors

Credits

Polling example

Download a voice changer file

Response

Errors

Credits

Example