Align takes audio and the text you already have for it (a script, captions, lyrics, an edited transcript) and returns your words, unchanged, each with a start and end time. Use it for karaoke-style highlighting, clickable transcripts and subtitle timing without retyping anything.
Words the audio doesn't clearly contain get times between their neighbors, and matched tells you how much of your text was heard.
Use it
curl https://labs.greatwork.company/v1/run \
-H "Authorization: Bearer $GWL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "align-1",
"input": "https://labs.greatwork.company/examples/voice-note.m4a",
"transcript": "Hey, it'\''s Sam. Quick update on the launch. The new pricing page is live, but the checkout button still points to the old plan. Can you fix that before three o'\''clock? Thanks."
}'const res = await fetch("https://labs.greatwork.company/v1/run", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.GWL_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "align-1",
"input": "https://labs.greatwork.company/examples/voice-note.m4a",
"transcript": "Hey, it's Sam. Quick update on the launch. The new pricing page is live, but the checkout button still points to the old plan. Can you fix that before three o'clock? Thanks."
}),
});
const { output, usage } = await res.json();import os, requests
res = requests.post(
"https://labs.greatwork.company/v1/run",
headers={"Authorization": f"Bearer {os.environ['GWL_API_KEY']}"},
json={
"model": "align-1",
"input": "https://labs.greatwork.company/examples/voice-note.m4a",
"transcript": "Hey, it's Sam. Quick update on the launch. The new pricing page is live, but the checkout button still points to the old plan. Can you fix that before three o'clock? Thanks."
},
)
output = res.json()["output"]# Send a local file instead of a URL
curl https://labs.greatwork.company/v1/run \
-H "Authorization: Bearer $GWL_API_KEY" \
-F model=align-1 \
-F input=@meeting.m4aSet GWL_API_KEY to a key from your dashboard. The bare name align works too and follows new versions.
Options
input: An audio or video file: an https URL, a data URL, or a file in a multipart upload, up to 25 MB and 5 minutes.
| Name | Type | Default | What it does |
|---|---|---|---|
transcript required | string | The words spoken in the audio, in order (up to 2,000 characters per call; split longer scripts with the audio). | |
language | string | Language of the speech (en, es...). Detected when omitted. |
Output
| Field | Type | Meaning |
|---|---|---|
words | object[] | Your words in order: text, start, end (seconds), matched (heard in the audio, or interpolated). |
matched | number | Share of your words found in the audio, 0 to 1. |
duration | number | Audio length in seconds. |
Every response wraps this in {"id", "model", "output", "usage": {"units", "unit", "cost_usd"}, "latency_ms"}.
Examples
Voice note script
{
"model": "align-1",
"input": "/examples/voice-note.m4a",
"transcript": "Hey, it's Sam. Quick update on the launch. The new pricing page is live, but the checkout button still points to the old plan. Can you fix that before three o'clock? Thanks."
}{
"words": [
{
"text": "Hey,",
"start": 0.24,
"end": 0.54,
"matched": true
},
{
"text": "it's",
"start": 0.68,
"end": 0.96,
"matched": true
},
{
"text": "Sam.",
"start": 0.96,
"end": 1.34,
"matched": true
},
{
"text": "Quick",
"start": 1.54,
"end": 1.84,
"matched": true
},
{
"text": "update",
"start": 1.84,
"end": 2.1,
"matched": true
},
{
"text": "on",
"start": 2.1,
"end": 2.24,
"matched": true
},
{
"text": "the",
"start": 2.24,
"end": 2.52,
"matched": true
},
{
"text": "launch.",
"start": 2.52,
"end": 2.9,
"matched": true
},
{
"text": "The",
"start": 3.04,
"end": 3.18,
"matched": true
},
{
"text": "new",
"start": 3.18,
"end": 3.56,
"matched": true
},
{
"text": "pricing",
"start": 3.56,
"end": 3.98,
"matched": true
},
{
"text": "page",
"start": 3.98,
"end": 4.14,
"matched": true
},
{
"text": "is",
"start": 4.14,
"end": 4.34,
"matched": true
},
{
"text": "live,",
"start": 4.34,
"end": 4.62,
"matched": true
},
{
"text": "but",
"start": 4.74,
"end": 4.9,
"matched": true
},
{
"text": "the",
"start": 4.9,
"end": 5.22,
"matched": true
},
{
"text": "checkout",
"start": 5.22,
"end": 5.66,
"matched": true
},
{
"text": "button",
"start": 5.66,
"end": 5.9,
"matched": true
},
{
"text": "still",
"start": 5.9,
"end": 6.24,
"matched": true
},
{
"text": "points",
"start": 6.24,
"end": 6.5,
"matched": true
},
{
"text": "to",
"start": 6.5,
"end": 6.66,
"matched": true
},
{
"text": "the",
"start": 6.66,
"end": 6.88,
"matched": true
},
{
"text": "old",
"start": 6.88,
"end": 7.24,
"matched": true
},
{
"text": "plan.",
"start": 7.24,
"end": 7.58,
"matched": true
},
{
"text": "Can",
"start": 7.8,
"end": 7.96,
"matched": true
},
{
"text": "you",
"start": 7.96,
"end": 8.24,
"matched": true
},
{
"text": "fix",
"start": 8.24,
"end": 8.44,
"matched": true
},
{
"text": "that",
"start": 8.44,
"end": 8.7,
"matched": true
},
{
"text": "before",
"start": 8.7,
"end": 9.08,
"matched": true
},
{
"text": "three",
"start": 9.08,
"end": 9.72,
"matched": false
},
{
"text": "o'clock?",
"start": 9.72,
"end": 10.36,
"matched": false
},
{
"text": "Thanks.",
"start": 10.36,
"end": 10.48,
"matched": true
}
],
"matched": 0.938,
"duration": 10.73
}Captured on our production servers: 7071 ms, 11 seconds, $0.000187.
Good for
- Highlight words as audio plays
- Time subtitles from an existing script
- Clickable, seekable transcripts
- Check a voiceover matches its script
Limits and caveats
- Your text should match what's said; a low
matchedmeans the script and audio differ. - Interpolated words get evenly spaced times, not measured ones.
Pricing
$0.001 per minute. Billed per second of audio, rounded up, at least one second. Failed calls aren't charged, and a call never costs more than its estimate (POST /v1/estimate tells you before you run it).
Built on
openai/whisper-base (timestamped) + our word alignment, 74M parameters. License: Apache-2.0. We run it on our servers.