One hour. One pass.
Preserve global context across continuous audio instead of stitching together disconnected chunks.
Production API for Microsoft VibeVoice-ASR
Long audio.
Clean answers.
A production API for Microsoft's VibeVoice-ASR—turning long-form, multilingual audio into speaker-aware transcripts with precise timestamps.
▶
customer-interview.wav42:18 · 3 speakers
TRANSCRIBED
Speaker 1
Let’s take the full recording—no manual slicing.
Speaker 2
And keep every speaker and timestamp attached.
Speaker 1
That’s exactly what this endpoint is built for.
60 minsingle pass
50+languages
Who · When · Whatstructured output
$0.0035per audio minute
01 / The product
Keep the context.
Lose the cleanup.
Built for the recordings that matter: research calls, podcasts, meetings, lectures, and interviews.
Speakers included.
Diarization and timestamps arrive with the transcript—not as a second job.
Speak naturally.
More than 50 languages, with native support for code-switching.
02 / How it works
01→
Create a key
Sign in with your email, add credit, generate an API key.
02→
Send audio
POST one continuous recording — raw bytes, up to 100 MiB.
03
Get structure
Poll the job and receive who spoke, when, and what they said.
Sign in with your email, create a key, top up a balance.
03 / API preview
Submit, poll,
collect the transcript.
One queue path for every length. Post your audio file, get a job id back in under a second, then poll until it's done — no timeouts on hour-long files.
SUBMIT
curl -X POST \
https://api.vibevoice-asr.com/v1/jobs \
-H "Authorization: Bearer $VVASR_KEY" \
-F file=@meeting.mp3
{ "job_id": "…", "status": "processing", "eta_seconds": 45 }GET /v1/jobs/{id}speaker · timestamps · text
04 / Pricing
PREPAID BALANCE
One price. No tiers.
Charged on the audio duration your file actually decodes to — not on upload size, not on wall-clock time. Failed and cancelled jobs cost nothing.
$0.0035per audio minute — a 60-minute recording is $0.21
Start with $5 ↗Long-form speech deserves long-form context.