$0.07 per audio-hour — dropping to $0.03 as you scale. Punctuated transcripts with per-word timestamps. English-specialized. We never store your audio — it's deleted the moment it's transcribed.
Get an API key See the APIBuilt to undercut the big STT vendors many-fold on English audio, with the margin coming from an efficient model — not corner-cutting.
A tiny REST API: POST an audio URL or upload, GET the transcript. Batch up to 100 files in one call. Webhooks if you'd rather not poll.
Native punctuation and casing, plus start/end for every word — ready for captions, search, and alignment.
Audio is processed and deleted on completion — no retention, never used for training. Early access, and a DPA is available on request.
Three calls. Point it at a URL (we fetch it) or upload a file.
# 1. submit a job curl -X POST https://api/v1/jobs \ -H "Authorization: Bearer $VERBII_KEY" -H "Content-Type: application/json" \ -d '{"audio_url":"https://your-bucket/call.mp3"}' # -> { "job_id": "job_abc", "status": "queued" } # 2. poll (or pass a callback_url for a webhook) curl https://api/v1/jobs/job_abc -H "Authorization: Bearer $VERBII_KEY" # -> { "status":"done", "text":"Hello there...", "words":[{"word":"Hello","start":0.0,"end":0.42}, ...] }
Every key includes 100 hours of audio free. Check your usage anytime at
GET /v1/usage and your jobs at GET /v1/jobs.
$0.07 per hour of audio — dropping to $0.03 as your volume grows. No minimums, no commitment.
Start with 100 free hours. The more you send, the less you pay.
Built for sensitive audio. The good stuff isn't a hidden enterprise toggle — it's the default.
Your audio is deleted the moment it's transcribed — by default, not a flag you have to find. Anything that fails to process is gone within 24 hours.
We don't train, fine-tune, or improve any model on your audio or transcripts. We don't sell or share it. Ever.
TLS in transit, AES-256 at rest, least-privilege access, private buckets. Transcripts auto-delete after 30 days (or sooner on request).
Processing in AWS (US); EU/UK in-region available on request. Read our Data Processing Agreement →
Straight talk on where we are: verbii is early access, built by a small team. Your audio is deleted on completion and never used for training — that part is real and on by default. But we are not HIPAA-eligible yet: no BAA, no SOC 2. So please don't send us PHI or other regulated data today. We'd rather tell you that up front than surprise you later.
Live check against the API. …
This page polls our health endpoint directly, so it reflects the real service — not a manually-updated badge.
We're early access, so we're honest about reliability: there's no blanket guarantee yet. A 99.5% uptime SLA — with service credits and a 4-business-hour response on outages — is available on request for committed customers, and we strengthen it as you scale, toward higher uptime and 24/7 support. Either way you get per-job status, clear error codes, and webhooks so you can see what's happening without waiting on us.
Paste a link or drop an audio file and see the transcript right here.
We're early and moving fast — here's what's live and what's landing next, so you can see the gaps closing.
Batch + URL/upload, per-word timestamps, per-word confidence, SRT/VTT subtitles, custom vocabulary (glossary), signed webhooks, idempotency, plus an llms.txt and an MCP server built for agents.
Who-said-what labels for meetings, calls, and interviews — our most-requested addition.
A web console and instant signup (we issue keys by hand today), plus an EU/UK data region on request.