Ave Public Captioning API

Base URL

https://api.el.ai/v1

Production

Authentication

Send a workspace API key as a bearer token on every request: Authorization: Bearer el_live_…. Read-only keys can read jobs, captions and the account; full keys can also upload and order. More on keys.

Endpoints

Jobs

Objects

The shapes the API sends and receives. Each endpoint page shows the ones it uses.

CreateJob

Provide exactly one source: media_url OR upload_id. Sending both or neither is invalid. Choose the Media URL or Uploaded file variant below.

media_url string

Public HTTP(S) video/audio URL, including supported YouTube links. Required only for the Media URL variant; omit when using upload_id.

≥ 1 chars
upload_id string · uuid

ID returned by POST /uploads, after uploading the file. Required only for the Uploaded file variant; omit when using media_url.
title string

Dashboard title and download filename. Defaults to the generated job ID. Default maximum: 300 characters, configurable per workspace.

≥ 1 chars
style string

clean removes fillers, false starts, and stutters; verbatim keeps every spoken word.

One of

clean
verbatim

Default

clean
model string

Captioning model. Only ave-base is supported.

One of

ave-base

Default

ave-base
speaker_labels boolean

Identify speaker turns. Set false for dialogue without labels.

Default

true
speaker_label_style string

SRT/VTT labels: >> NAME:, [NAME], NAME:, or dashes without names. Ignored when speaker_labels is false. Transcripts use their native speaker headings; JSON has a separate speaker field.

One of

gt_name
brackets
colon
dashes

Default

gt_name
atmospherics boolean

Include meaningful non-speech sounds. Uses the same add-on pricing as dashboard checkout.

Default

true
profanity string

Keep profanity, mask it (f***), replace it with [expletive], or omit it. Does not alter source audio.

One of

as_spoken
mask
label
omit

Default

as_spoken
redactions Redactions

Privacy controls for caption text, using the same redaction pricing as checkout. Source audio and video are unchanged.
Accepts one of:
- Standard replacement
- Custom replacement requires replacement, custom_replacement
- pii boolean
  
  Redact personal data such as emails and phone numbers.
  
  Default
  
  false
- names boolean
  
  Redact named individuals.
  
  Default
  
  false
- organizations boolean
  
  Redact organization names.
  
  Default
  
  false
- terms array of string
  
  Custom terms to redact, one term per item; each must be a nonblank single line.
  
  Default
  
  []
  
  ≤ 200 items
- replacement string
  
  Replace with [REDACTED], a category such as [EMAIL], nothing, or custom_replacement.
  
  One of
  
  redacted
  category_label
  silent
  custom_string
  
  Default
  
  category_label
- custom_replacement string
  
  Required when replacement is custom_string. Nonblank, single-line text.
  
  1–120 chars
glossary array of string

Names and terms with preferred spelling, used where the audio supports them. Defaults to at most 200 terms of 100 characters each; configurable per workspace.

Default

[]
webhook_url string · uri

Public HTTPS callback. Redirects are not followed.

≤ 2048 chars
matches ^https://
metadata map of string

Your string key–value pairs, returned on the job and in webhook events. Defaults: 20 keys of 1–40 characters, values up to 500 characters. These maxima are configurable per workspace.

Job

id string · uuid
object string

One of

job
status string

One of

queued
processing
completed
cancelled
progress integer · nullable

0–100
eta string · date-time · nullable
title string
media object
- type string
  
  One of
  
  url
  upload
- url string · nullable
- upload_id string · uuid · nullable
- duration_seconds number
language string · nullable
options object
- style string
  
  clean removes fillers, false starts, and stutters; verbatim keeps every spoken word.
  
  One of
  
  clean
  verbatim
  
  Default
  
  clean
- model string
  
  Captioning model. Only ave-base is supported.
  
  One of
  
  ave-base
  
  Default
  
  ave-base
- speaker_labels boolean
  
  Identify speaker turns. Set false for dialogue without labels.
  
  Default
  
  true
- speaker_label_style string
  
  SRT/VTT labels: >> NAME:, [NAME], NAME:, or dashes without names. Ignored when speaker_labels is false. Transcripts use their native speaker headings; JSON has a separate speaker field.
  
  One of
  
  gt_name
  brackets
  colon
  dashes
  
  Default
  
  gt_name
- atmospherics boolean
  
  Include meaningful non-speech sounds. Uses the same add-on pricing as dashboard checkout.
  
  Default
  
  true
- profanity string
  
  Keep profanity, mask it (f***), replace it with [expletive], or omit it. Does not alter source audio.
  
  One of
  
  as_spoken
  mask
  label
  omit
  
  Default
  
  as_spoken
- redactions Redactions
  
  Privacy controls for caption text, using the same redaction pricing as checkout. Source audio and video are unchanged.
  Accepts one of:
  Standard replacement
  Custom replacement requires replacement, custom_replacement
  pii boolean
  
  Redact personal data such as emails and phone numbers.
  
  Default
  
  false
  
  names boolean
  
  Redact named individuals.
  
  Default
  
  false
  
  organizations boolean
  
  Redact organization names.
  
  Default
  
  false
  
  terms array of string
  
  Custom terms to redact, one term per item; each must be a nonblank single line.
  
  Default
  
  []
  
  ≤ 200 items
  
  replacement string
  
  Replace with [REDACTED], a category such as [EMAIL], nothing, or custom_replacement.
  
  One of
  
  redacted
  category_label
  silent
  custom_string
  
  Default
  
  category_label
  
  custom_replacement string
  
  Required when replacement is custom_string. Nonblank, single-line text.
  
  1–120 chars
- glossary array of string
- webhook_url string · nullable
cost object · nullable
- amount string
  
  Decimal amount in USD, net of refunds.
- currency string
  
  One of
  
  usd
cancellation object · nullable
- reason string
  
  One of
  
  unsupported_language
  unsupported_media
  unusable_media
  media_unavailable
  deleted
  not_started
- message string
captions object · nullable
- srt string · uri
- vtt string · uri
- txt string · uri
- docx string · uri
- json string · uri
metadata map of string

String values. Defaults: 20 keys of 1–40 characters, values up to 500 characters. These maxima are configurable per workspace.
created_at string · date-time
completed_at string · date-time · nullable

Error

error object
- code string
- message string
- param string

Upload

id string · uuid
object string

One of

upload
upload_url string · uri

Presigned URL for uploading raw file bytes with PUT. No Authorization or storage-specific headers are needed.
method string

One of

PUT
max_size_bytes integer

One of

5242880000
expires_at string · date-time

Account

object string

One of

account
workspace object
- id string · uuid
- name string
key object
- id string · uuid
- name string
- access string
  
  One of
  
  full
  read_only
billing object
- mode string · nullable
  
  One of
  
  prepaid
  invoiced
- balance string · nullable
- currency string
  
  One of
  
  usd
- rate_per_minute string · nullable

Event

id string

matches ^evt_
object string

One of

event
type string

One of

job.completed
job.cancelled
created_at string · date-time
data Job
- id string · uuid
- object string
  
  One of
  
  job
- status string
  
  One of
  
  queued
  processing
  completed
  cancelled
- progress integer · nullable
  
  0–100
- eta string · date-time · nullable
- title string
- media object
  type string
  
  One of
  
  url
  upload
  
  url string · nullable
  
  upload_id string · uuid · nullable
  
  duration_seconds number
- language string · nullable
- options object
  style string
  
  clean removes fillers, false starts, and stutters; verbatim keeps every spoken word.
  
  One of
  
  clean
  verbatim
  
  Default
  
  clean
  
  model string
  
  Captioning model. Only ave-base is supported.
  
  One of
  
  ave-base
  
  Default
  
  ave-base
  
  speaker_labels boolean
  
  Identify speaker turns. Set false for dialogue without labels.
  
  Default
  
  true
  
  speaker_label_style string
  
  SRT/VTT labels: >> NAME:, [NAME], NAME:, or dashes without names. Ignored when speaker_labels is false. Transcripts use their native speaker headings; JSON has a separate speaker field.
  
  One of
  
  gt_name
  brackets
  colon
  dashes
  
  Default
  
  gt_name
  
  atmospherics boolean
  
  Include meaningful non-speech sounds. Uses the same add-on pricing as dashboard checkout.
  
  Default
  
  true
  
  profanity string
  
  Keep profanity, mask it (f***), replace it with [expletive], or omit it. Does not alter source audio.
  
  One of
  
  as_spoken
  mask
  label
  omit
  
  Default
  
  as_spoken
  
  redactions Redactions
  
  Privacy controls for caption text, using the same redaction pricing as checkout. Source audio and video are unchanged.
  
  Accepts one of:
  Standard replacement
  Custom replacement requires replacement, custom_replacement
  
  pii boolean
  
  Redact personal data such as emails and phone numbers.
  
  Default
  
  false
  
  names boolean
  
  Redact named individuals.
  
  Default
  
  false
  
  organizations boolean
  
  Redact organization names.
  
  Default
  
  false
  
  terms array of string
  
  Custom terms to redact, one term per item; each must be a nonblank single line.
  
  Default
  
  []
  
  ≤ 200 items
  
  replacement string
  
  Replace with [REDACTED], a category such as [EMAIL], nothing, or custom_replacement.
  
  One of
  
  redacted
  category_label
  silent
  custom_string
  
  Default
  
  category_label
  
  custom_replacement string
  
  Required when replacement is custom_string. Nonblank, single-line text.
  
  1–120 chars
  
  glossary array of string
  
  webhook_url string · nullable
- cost object · nullable
  amount string
  
  Decimal amount in USD, net of refunds.
  
  currency string
  
  One of
  
  usd
- cancellation object · nullable
  reason string
  
  One of
  
  unsupported_language
  unsupported_media
  unusable_media
  media_unavailable
  deleted
  not_started
  
  message string
- captions object · nullable
  srt string · uri
  
  vtt string · uri
  
  txt string · uri
  
  docx string · uri
  
  json string · uri
- metadata map of string
  
  String values. Defaults: 20 keys of 1–40 characters, values up to 500 characters. These maxima are configurable per workspace.
- created_at string · date-time
- completed_at string · date-time · nullable

Redactions

Privacy controls for caption text, using the same redaction pricing as checkout. Source audio and video are unchanged.

pii boolean

Redact personal data such as emails and phone numbers.

Default

false
names boolean

Redact named individuals.

Default

false
organizations boolean

Redact organization names.

Default

false
terms array of string

Custom terms to redact, one term per item; each must be a nonblank single line.

Default

[]

≤ 200 items
replacement string

Replace with [REDACTED], a category such as [EMAIL], nothing, or custom_replacement.

One of

redacted
category_label
silent
custom_string

Default

category_label
custom_replacement string

Required when replacement is custom_string. Nonblank, single-line text.

1–120 chars

Ave Public Captioning API

Base URL

Authentication

Endpoints

Jobs

Uploads

Account

Webhooks

Objects

CreateJob

Job

Error

Upload

Account

Event

Redactions