Meeting AI

Upload a meeting recording, poll until processing finishes, then read the speaker-labelled transcript, summary, decisions, and action items.

Authentication

Use a server-side developer API key with Meeting AI → Access, permission ID meetings. Access grants meetings:read and meetings:write. Restricted keys need this permission added explicitly. See API keys.

Credential	Route prefix	Purpose
Developer API key	/v1/meetings	Server integrations documented here
User access JWT	/api/v1/meetings	VoiceLab dashboard

Meeting AI is available to personal workspace keys only. Team workspace keys return 403 workspace_product_unavailable. Meetings belong to the account: a permitted key can retrieve meetings created by another key or in the web app, and meetings created with a key appear in the web Meeting AI list. Other accounts cannot access them; their IDs return 404 meeting_not_found.

Endpoints

Method	Path	Scope	Purpose
GET	/v1/meetings/capabilities	meetings:read	Supported languages and feature flags
POST	/v1/meetings	meetings:write	Upload a recording; returns 202
GET	/v1/meetings	meetings:read	List meetings or refresh selected IDs
GET	/v1/meetings/{id}	meetings:read	Poll status and read results
PATCH	/v1/meetings/{id}	meetings:write	Rename a meeting
DELETE	/v1/meetings/{id}	meetings:write	Delete a meeting
POST	/v1/meetings/{id}/notes/generations	meetings:write	Regenerate notes after a notes failure

Developer responses never include audio playback URLs. Audio download, transcript and notes editing, speaker renaming, translations, and webhooks are not available with API keys. Meeting AI has no completion webhooks; poll instead.

Check capabilities

GEThttps://api.voicelab.uz/v1/meetings/capabilities

List supported recording languages before uploading.

Requires meetings:read.

curl --fail-with-body 'https://api.voicelab.uz/v1/meetings/capabilities' \
  -H "Authorization: Bearer $VOICELAB_API_KEY"

Returns 200. data lists supported languages with code, name, and native_name. Send one code as language, or omit it for automatic detection. Allow unknown fields in this response.

Create a meeting

POSThttps://api.voicelab.uz/v1/meetings

Upload a recording for asynchronous transcription and notes.

Requires meetings:write.

Headers and fields

Header	Required	Value
Authorization	yes	Bearer <VOICELAB_API_KEY>
Idempotency-Key	yes	UUID generated once per logical upload; reuse for retries
Content-Type	yes	Multipart boundary generated by your HTTP client

Multipart field	Required	Value
audio	yes	One audio file
title	no	1–160 characters, single line; defaults to the filename without its extension
language	no	auto or a code from capabilities; default auto

Send each field at most once. Unknown or repeated parts return 400 invalid_multipart. API-key uploads are always recorded as uploads; the web app's source field is not accepted.

The same idempotency key and content return the original meeting with Idempotent-Replayed: true and no second credit hold. Changing the audio, title, or language with that key returns 409 idempotency_key_reused. Keys are scoped per developer API key, separately from web app uploads.

Limits

Limit	Value
Upload size	500 MiB, or 524,288,000 bytes
Duration	0.5 seconds to 24 hours
Formats	MP3, WAV, M4A/AAC, OGG/Opus, WebM/Opus, or FLAC
Pending meetings	10 queued or running meetings per account
Write rate	Shared per-account budget for uploads and notes regeneration

The API inspects the file contents and ignores the client-supplied content type. Write requests share their per-account rate limit with the Meeting AI web app, so dashboard activity and API traffic draw from the same budget. Reads have a separate per-account budget. When ten meetings are already queued or running, new uploads return 429 meeting_queue_full; wait for one to finish. Respect Retry-After on every 429.

Request examples

These examples run on your server. Set VOICELAB_API_KEY in the environment. Let the client generate the multipart boundary. Retain the generated UUID if you retry the upload.

cURL

request_key=$(uuidgen)
curl --fail-with-body -i 'https://api.voicelab.uz/v1/meetings' \
  -H "Authorization: Bearer $VOICELAB_API_KEY" \
  -H "Idempotency-Key: $request_key" \
  -F 'audio=@weekly-sync.mp3' \
  -F 'title=Weekly sync' \
  -F 'language=en'

Python

import os
import time
import uuid
import requests

api = "https://api.voicelab.uz/v1/meetings"
headers = {"Authorization": f"Bearer {os.environ['VOICELAB_API_KEY']}"}
request_key = str(uuid.uuid4())

with open("weekly-sync.mp3", "rb") as audio:
    response = requests.post(
        api,
        headers={**headers, "Idempotency-Key": request_key},
        files={"audio": ("weekly-sync.mp3", audio, "audio/mpeg")},
        data={"title": "Weekly sync", "language": "en"},
        timeout=900,
    )
response.raise_for_status()
meeting = response.json()["meeting"]

while meeting["status"] in ("queued", "running"):
    time.sleep(int(response.headers.get("Retry-After", "2")))
    response = requests.get(f"{api}/{meeting['id']}", headers=headers, timeout=30)
    response.raise_for_status()
    meeting = response.json()["meeting"]

if meeting["status"] == "failed":
    raise RuntimeError(meeting["failure"]["code"])
print(meeting["transcript"]["text"])
print(meeting["notes"]["summary"])

TypeScript

import { readFile } from "node:fs/promises";
import { randomUUID } from "node:crypto";

const requestKey = randomUUID();
const form = new FormData();
form.append("audio", new Blob([new Uint8Array(await readFile("weekly-sync.mp3"))]), "weekly-sync.mp3");
form.append("title", "Weekly sync");
form.append("language", "en");

const response = await fetch("https://api.voicelab.uz/v1/meetings", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.VOICELAB_API_KEY}`,
    "Idempotency-Key": requestKey,
  },
  body: form,
  signal: AbortSignal.timeout(900_000),
});
if (!response.ok) throw new Error(`Upload failed: ${response.status}`);
console.log((await response.json()).meeting.id);

Response and polling

Returns 202 Accepted, Location: /v1/meetings/{id}, and Retry-After: 2. Retries with the same key and content return the existing meeting:

{
  "meeting": {
    "id": "mtg_550e8400-e29b-41d4-a716-446655440000",
    "title": "Weekly sync",
    "status": "queued",
    "capture_source": "upload",
    "created_at": "2026-10-14T09:00:00Z",
    "updated_at": "2026-10-14T09:00:00Z",
    "audio": {
      "filename": "weekly-sync.mp3",
      "content_type": "audio/mpeg",
      "size_bytes": 7340032,
      "duration_ms": 1830000
    },
    "transcript": { "text": "", "language": "", "language_probability": 0, "segments": [] },
    "notes": { "version": 0, "summary": "", "decisions": [], "action_items": [] },
    "versions": [],
    "processing": {
      "transcription": { "status": "queued", "retryable": false },
      "notes": { "status": "waiting", "retryable": false }
    }
  },
  "async": true,
  "request_id": "req_example"
}

Poll GET /v1/meetings/{id} every two seconds, or follow Retry-After, until status is no longer queued or running:

status	Meaning	Next step
queued	Waiting for the processor	Keep polling
running	Transcription or notes in progress	Keep polling
transcribed	Transcript saved; notes generation failed	Read the transcript; optionally regenerate notes
ready	Transcript and notes complete	Read results
failed	Processing failed; no transcript	Inspect failure.code; upload again if appropriate

processing.transcription and processing.notes report each stage separately: waiting, queued, running, ready, or failed, with a retryable flag and a safe failure object when a stage fails. A meeting can return to queued or running with its transcript intact while notes are regenerated.

When the meeting is ready, read these fields:

Field	Meaning
transcript.text	Full transcript text
transcript.language, transcript.language_probability	Detected or requested language and confidence
transcript.segments	Timed segments with speaker_id, speaker, text, and words
notes.summary	Meeting summary
notes.decisions	Decisions, each with evidence.segment_ids
notes.action_items	Action items with owner and evidence.segment_ids
failure	code and message when processing fails

List meetings

GEThttps://api.voicelab.uz/v1/meetings

List the account's meetings, newest first.

Requires meetings:read. Optional limit defaults to 20 and accepts 1–50. Pass the returned opaque next_cursor as cursor for the next page. To refresh specific meetings, send ids as a comma-separated list of up to 50 meeting IDs instead of cursor. IDs that do not belong to the account are omitted.

curl --fail-with-body 'https://api.voicelab.uz/v1/meetings?limit=20' \
  -H "Authorization: Bearer $VOICELAB_API_KEY"

Returns 200 with meetings in descending creation order and optional next_cursor. Invalid limit, cursor, or ids values return 400 invalid_meeting_query.

Get a meeting

GEThttps://api.voicelab.uz/v1/meetings/{id}

Read meeting status, transcript, and notes.

Requires meetings:read. Returns 200 with meeting. Unknown meetings, deleted meetings, and meetings belonging to another account return 404 meeting_not_found.

curl --fail-with-body 'https://api.voicelab.uz/v1/meetings/mtg_REPLACE_ME' \
  -H "Authorization: Bearer $VOICELAB_API_KEY"

Rename a meeting

PATCHhttps://api.voicelab.uz/v1/meetings/{id}

Change a meeting's title.

Requires meetings:write. Send only title, 1–160 characters on one line. Unknown fields return 422 validation_error.

curl --fail-with-body -X PATCH 'https://api.voicelab.uz/v1/meetings/mtg_REPLACE_ME' \
  -H "Authorization: Bearer $VOICELAB_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"title":"Weekly sync — product"}'

Returns 200 with the updated meeting.

Delete a meeting

DELETEhttps://api.voicelab.uz/v1/meetings/{id}

Delete a meeting and its results.

Requires meetings:write. Returns 204 No Content. Deleting a queued or running meeting releases its credit hold. Deleted meetings also disappear from the web app.

curl --fail-with-body -X DELETE 'https://api.voicelab.uz/v1/meetings/mtg_REPLACE_ME' \
  -H "Authorization: Bearer $VOICELAB_API_KEY"

Regenerate notes

POSThttps://api.voicelab.uz/v1/meetings/{id}/notes/generations

Retry notes for a meeting whose notes generation failed.

Requires meetings:write. Only meetings with status transcribed qualify: the transcript exists and notes generation failed. Other meetings return 409 meeting_notes_not_retryable. Regeneration is billed again at the meeting's full duration and counts toward the key's credit limit and the shared write rate limit.

curl --fail-with-body -X POST 'https://api.voicelab.uz/v1/meetings/mtg_REPLACE_ME/notes/generations' \
  -H "Authorization: Bearer $VOICELAB_API_KEY"

Returns 202 Accepted with meeting, Location: /v1/meetings/{id}, and Retry-After: 2. Poll the meeting as after an upload.

Billing and errors

Credits are reserved when the upload is accepted, prorated by recording duration at the configured per-minute rate and rounded up to a whole credit. They are debited once when processing succeeds. Processing failures and deletion before completion release the hold. Polling, listing, renaming, and deleting are not charged. Notes regeneration reserves and charges again. See pricing.

A key's credit limit includes its recorded billable usage and pending reservations across Meeting AI, Voice Isolator, and LLM requests. Uploads or regenerations that would exceed it return 402 api_key_credit_limit. Revoking a key prevents new access but does not cancel accepted meetings. Usage is attributed to the submitting key under meeting_ai in analytics.

HTTP status	Code	Meaning
400	invalid_idempotency_key	Missing or non-UUID Idempotency-Key
400	invalid_multipart	Malformed form, missing audio, or unknown/repeated parts
400	invalid_meeting_query	Invalid limit, cursor, or ids
401	invalid_api_key	Missing, invalid, expired, disabled or revoked API key
402	insufficient_credits	Account balance cannot cover the reservation
402	api_key_credit_limit	Request would exceed the key's credit limit
403	insufficient_scope	Key lacks meetings:read or meetings:write
403	workspace_product_unavailable	Key belongs to a team workspace
403	workspace_mismatch, workspace_forbidden	Workspace header or membership does not match the key
404	meeting_not_found	Unknown, deleted, or another account's meeting
409	idempotency_key_reused	Same key sent with different content
409	meeting_processing	The original request with this key is still in progress
409	meeting_notes_not_retryable	Notes regeneration requested for a meeting that is not transcribed
409	meeting_not_ready	Operation requires completed processing
413	audio_too_large	Upload exceeds 500 MiB
415	unsupported_audio_format	File contents are not a supported audio format
422	invalid_audio	Unreadable audio, or duration outside 0.5 seconds–24 hours
422	invalid_title, invalid_language, validation_error	Invalid title, language, or rename body
429	rate_limited	Shared per-account request budget exceeded
429	meeting_queue_full	Ten meetings are already queued or running
429	meeting_upload_capacity	Upload capacity is temporarily busy
503	meeting_billing_unavailable	Pricing unavailable; nothing was queued
503	meeting_feature_unavailable, meeting_worker_unavailable	Meeting AI or its processor is unavailable; nothing was queued
503	developer_auth_unavailable	API key verification is temporarily unavailable
504	meeting_request_timeout	Request timed out; check status before retrying
500	meeting_internal_error	Unexpected failure

Honor Retry-After for 429 and 503 responses. Retry transient upload failures with the original idempotency key and identical content; after a timeout or lost response, the retry returns the existing meeting instead of creating a duplicate. Do not retry authorization or validation failures unchanged. Keep request_id for support. See errors and limits.