99 languages, no setup
The language is detected automatically: English, Spanish, Polish, Russian and dozens more in a single stream.
service is live
Audio transcription to text
via API
Turn calls, voice messages and interviews into text in 99 languages. The protocol matches OpenAI Whisper, so switching takes one line of code.
10 minutes free — right after sign-up
After that — up to 0.006 USD per minute of audio, pay as you go
Why teams choose voicescribe
Results in seconds
A minute of audio is transcribed in a couple of seconds and returned in the same request — no queues, no polling.
Data does not pile up
Recordings and transcripts live no longer than 24 hours — just long enough to verify quality — then they are deleted.
Keep your code
Works with the official openai-sdk: only the base URL and key change. Response and error formats are identical.
Connect in a minute
Already calling OpenAI Whisper? Replace two lines.
from openai import OpenAI
client = OpenAI(base_url="https://voicesscribe.com/v1", api_key="your key")
r = client.audio.transcriptions.create(model="whisper-1", file=open("call.ogg", "rb"))
print(r.text)1
Get a key
Sign-up takes a minute; the key is issued instantly along with client area access.
2
Change the base URL
Point the SDK at our endpoint with your key — the rest of your code stays untouched.
3
Track usage
Minutes, transcript history and the invoice for the period are in your client area, live.
Speech recognition API features
A single endpoint POST /v1/audio/transcriptions — just like OpenAI.
Audio formats
ogg · opus · mp3 · wav · m4a · webm — up to 25 MB
Response formats
json · text · verbose_json · srt · vtt
Term hints
The prompt parameter improves accuracy for names, brands and codes.
Subtitles and timings
Ready-made srt and vtt, or timed segments in verbose_json.
Guides and examples
Step-by-step instructions for the usual tasks — with code and error handling.
Transcribing call recordings
How to turn phone conversations into text and put it on a conveyor.
Meeting recordings to notes
Zoom, Teams and Meet recordings into searchable text and action items.
Automatic SRT and VTT subtitles
Ready-made subtitles from a webinar or a video in a single request.
Improving accuracy
Eight fixes that actually move the word error rate on your own audio.
Frequently asked questions
How is it different from the OpenAI Whisper API?
Same protocol and response formats, so your code stays as is — just point it to a different base URL and key. Recognition runs on our own hardware.
Which languages are supported?
99 languages including English, Spanish, German, Polish and Russian. The language is detected automatically, specifying it is optional.
How fast is transcription?
One minute of audio takes a couple of seconds. The text comes back in the same request — no status polling required.
What happens to uploaded recordings?
Recordings and transcripts are kept for no longer than 24 hours so you can verify quality, then deleted automatically.
Which audio formats are accepted?
ogg, opus, mp3, wav, m4a, webm and other common formats up to 25 MB. Responses are available as json, text, verbose_json, srt and vtt.
How much does it cost?
Billing is per minute of audio: up to 0.006 USD per minute, with volume pricing on request. Only successful requests are counted — silence and errors are not billed.