Audio transcription API with speakers and word timestamps
Send the URL of an audio or video file and get the transcript back, with speaker labels, a timestamp on every word and the language detected automatically. The rate is 6 credits per minute of audio.
{
"success": true,
"data": {
"audio_url": "https://files.example.com/calls/2026-11-02-discovery.mp3",
"language": "en",
"duration_seconds": 1260,
"speaker_count": 2,
"text": "Thanks for joining, Priya. Can you walk me through how inbound leads are handled today? Sure. Every form fill lands in a shared inbox, and someone calls the number by hand.",
"segments": [
{
"speaker": "A",
"start": 0.48,
"end": 5.12,
"text": "Thanks for joining, Priya. Can you walk me through how inbound leads are handled today?"
},
{
"speaker": "B",
"start": 5.74,
"end": 11.3,
"text": "Sure. Every form fill lands in a shared inbox, and someone calls the number by hand."
}
],
"words": [
{
"word": "Thanks",
"start": 0.48,
"end": 0.81,
"speaker": "A"
},
{
"word": "for",
"start": 0.81,
"end": 0.95,
"speaker": "A"
},
{
"word": "joining,",
"start": 0.95,
"end": 1.42,
"speaker": "A"
},
{
"word": "Priya.",
"start": 1.47,
"end": 1.98,
"speaker": "A"
}
]
},
"meta": {
"credits_used": 6,
"credits_remaining": 999,
"cached": false,
"request_id": "req_8f3a2c1e9b7d"
}
}What you get back
The fields most teams branch on. The full response is in the sample on this page and in the API reference.
textTranscriptThe full transcript as a single string.segmentsSpeaker segmentsThe transcript split into turns, each with a speaker label and start and end times.wordsWord timestampsEvery word with its start and end time in seconds.languageDetected languageThe language code identified from the audio.speaker_countSpeaker countHow many distinct speakers were heard.duration_secondsDurationLength of the audio, which is what billing is based on.What Audio Transcription does
Audio or video by URL
Point the API at a file that your storage or call platform already hosts. There is no upload step, and video files are transcribed from their audio track.
Speaker labels on by default
Turns are attributed to speakers without any setup, so a two-person call reads as a conversation. Set speakers to off when you only need the words.
Word-level timestamps
Each word carries a start and an end time. Jump a player to the moment a phrase was said, cut clips, or build captions.
Automatic language detection
The language is detected from the audio across 99 supported languages. You do not need to tag recordings before you send them.
See the response before you sign up
Fill in the request and run it. The dashboard has the same form, plus history, bulk upload and live data.
Fill in the field and run the lookup to see what comes back.
One POST request
Send JSON with a Bearer key. The same key works for all 79 products.
curl -X POST https://api.1lookup.dev/v1/audio-transcription \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"audio_url":"https://files.example.com/calls/2026-11-02-discovery.mp3","speakers":"on"}'- Endpoint
POST /v1/audio-transcription- Base URL
https://api.1lookup.dev- Request
audio_urlspeakers- Price
- 6 credits / minute
- Coverage
- 99 languages, automatic detection
- Bulk
- CSV upload and bulk API
- Cache
- Repeats within 7 days are free
- Also via
- Dashboard and MCP server (run_lookup)
What teams use it for
Review sales and support calls
Transcribe recordings in bulk, then search them for competitor mentions, objections or required disclosures.
Summarize meetings with an LLM
Send the speaker-labeled transcript to a model for notes and action items that say who committed to what.
Caption video and podcasts
Turn word timestamps into subtitle files and a searchable transcript page for each episode.
Run call center quality checks
Put each day's recordings through the bulk API and score scripts and disclosure language on text, instead of sampling calls by ear.
Recording and transcribing calls can require consent from the people on them, depending on where they are. Make sure you have the right to process each recording. This note is informational and is not legal advice.
6 credits / minute
Credits come with your plan and work on every product. Requests that fail on our side are never charged. Beyond your plan, extra credits are pay as you go.
See pricing| Plan | Per 1,000 credits | Per 1,000 minutes |
|---|---|---|
| Starter $29/mo | $4.83 | $29.00 |
| Growth $99/mo | $3.96 | $23.76 |
| Pro $299/mo | $3.32 | $19.93 |
| Scale $799/mo | $2.66 | $15.98 |
Audio Transcription questions
What do I send to the transcription API?
A JSON body with audio_url, the link to an audio or video file, and optionally speakers set to on or off. The file has to be reachable at that URL when the request is made.
How is audio transcription billed?
6 credits per minute of audio, billed per minute. A 30-minute recording costs 180 credits. Speaker labels and word timestamps are included in that price.
Which languages can it transcribe?
99 languages, detected automatically, so you do not pass a language code. Quality depends on the recording: clear speech transcribes better than crosstalk, heavy background noise or a poor phone line.
How do speaker labels work?
Speakers are separated by voice and labeled in order of appearance, such as A and B. Labels are not names. The API does not know who is speaking, so map labels to people in your own system.
Can I transcribe call recordings in bulk?
Yes. Submit a list of URLs to the bulk API or upload a CSV. The job takes an idempotency key, reports errors per row and fires a webhook when the transcripts are ready.
Related products
Video Transcript
The spoken-word transcript of a public video on TikTok, Instagram, YouTube, X or Facebook.
Website Scraper
Any URL as clean Markdown and metadata, ready for LLM and RAG pipelines.
Phone Validation
Validity, carrier, line type, porting, DNC status and a risk level for any number.
Run Audio Transcription on your own data
Starter is $29 a month for 6,000 credits, enough for 1,000 minutes a month.