Retrieve an Edit

The Python SDK handles this automatically. Use process() — it submits the job, polls until completion, and returns the result. You don't need to call get_edit() directly.

result = client.process("https://example.com/episode.mp3", fillers=True)
result.audio.download("cleaned.mp3")

Legacy: manual polling with `get_edit()`

Use this only if you submitted a job via create_edit() and need to check its status separately — for example in a background worker or webhook handler.

edit = client.get_edit("edit_abc123")
print(edit.status)  # PENDING, PREPROCESSING, CLASSIFICATION, EDITING, POSTPROCESSING, EXPORT, SUCCESS, FAILURE, RETRY

if edit.status == "SUCCESS":
    print(edit.result.download_url)

The JavaScript SDK handles this automatically. Use process() — it submits the job, polls until completion, and resolves with the result. You don't need to call getEdit() directly.

const result = await client.process('https://example.com/episode.mp3', { fillers: true });
console.log(result.audio.url);

Legacy: manual polling with `getEdit()`

Use this only if you submitted a job via createEdit() and need to check its status separately.

const edit = await client.getEdit('edit_abc123');
console.log(edit.status); // PENDING, PREPROCESSING, CLASSIFICATION, EDITING, POSTPROCESSING, EXPORT, SUCCESS, FAILURE, RETRY

if (edit.status === 'SUCCESS' && edit.result && 'download_url' in edit.result) {
  console.log(edit.result.download_url);
}

getEdit() / get_edit() returns the raw API polling payload with fields like download_url, transcription, summarization, and social_content. The higher-level process() helpers reshape that into SDK-friendly fields like result.audio.url and result.transcript.

A recent live social_content=true check returned a populated social_content object with newsletter, twitter_thread, and linkedin, and remained in POSTPROCESSING for several polls before SUCCESS.

Endpoint

GET /v2/edits/{edit_id}

Poll this endpoint after submitting a job until status is SUCCESS or FAILURE.

curl https://api.cleanvoice.ai/v2/edits/edit_abc123 \
  -H "X-API-Key: $CLEANVOICE_API_KEY"

Response

{
  "task_id": "edit_abc123",
  "status": "SUCCESS",
  "result": {
    "download_url": "https://storage.cleanvoice.ai/cleaned/episode.mp3",
    "transcription": null,
    "summarization": null,
    "social_content": []
  }
}

Real successful responses are usually larger than this. In live API calls, result.transcription contains paragraph and word timing data, result.summarization contains title/summary/chapter fields, and result.social_content is an object with newsletter, twitter_thread, and linkedin when social content generation is enabled. See the REST retrieve page for trimmed anonymized examples of those full payloads.

Status values

Status	Description
`PENDING`	Job is queued
`STARTED`	Processing in progress
`PREPROCESSING`	Input analysis and setup is in progress
`CLASSIFICATION`	Cleanvoice is classifying events in the media
`EDITING`	The main edit pass is running
`POSTPROCESSING`	Final cleanup and result assembly is running
`EXPORT`	Output files are being written
`SUCCESS`	Done — result is available
`FAILURE`	Processing failed
`RETRY`	Temporary failure, retrying automatically

Polling loop

while true; do
  RESPONSE=$(curl -s https://api.cleanvoice.ai/v2/edits/edit_abc123 \
    -H "X-API-Key: $CLEANVOICE_API_KEY")
  STATUS=$(echo $RESPONSE | jq -r '.status')
  echo "Status: $STATUS"
  if [ "$STATUS" = "SUCCESS" ] || [ "$STATUS" = "FAILURE" ]; then
    echo $RESPONSE
    break
  fi
  sleep 10
done

Start polling after ~30 seconds. Poll every 10 seconds after that. Processing typically takes ~30s for short clips, 5–10 min for 1h files.

Legacy: manual polling with get_edit()