What Is WhisperAPI and What Does It Do?

WhisperAPI is a hosted transcription service that turns audio and video files into text using OpenAI's Whisper speech recognition models. It fits two kinds of users: developers who want to add transcription to an application through a REST endpoint, and non-developers who just want to upload a file and get a transcript back. You can start with 5 free credits and no credit card, then pay as you go.

What problem it solves

Running Whisper yourself means provisioning a GPU, managing model weights, and handling long-file chunking. WhisperAPI moves that work behind an HTTP call or a drag-and-drop dashboard, so you send a file and receive text in a chosen format. The site reports 98+ supported languages, file uploads up to 10GB, no minute limits, and results "in minutes, not hours" — its own example is 10 minutes of audio transcribed in under a minute.

Two ways to use it

For developers: the REST API

The documented request is a curl POST to https://api.whisper-api.com/transcribe with an API key header and multipart form fields:

curl \
  -F "[email protected]" \
  -F "language=en" \
  -F "format=srt" \
  -F "model_size=large-v2" \
  -H "X-API-Key: YOUR_API_KEY" \
  https://api.whisper-api.com/transcribe

What each input does:

• file — the audio or video file to transcribe (the same endpoint accepts both).
• language — the spoken language, e.g. en.
• format — output format; the site lists JSON, TEXT, VTT, SRT, DOCX, and PDF.
• model_size — which Whisper model to run, trading speed against accuracy (the site names large-v2 in its example).
• X-API-Key — your credential.

The site also states the API supports remote URLs as an alternative to direct uploads and lets you fine-tune model parameters. Expected result: a transcription response in the format you requested.

For everyone else: the no-code dashboard

The dashboard offers drag-and-drop upload, real-time progress, multi-format download, and a single place to manage past transcriptions. No API integration is required.

Key facts at a glance

How to decide if it fits

• Choose the API if transcription is one step inside a larger pipeline — for example, a service that ingests user-uploaded recordings and returns subtitles. You get model selection and parameter control without running infrastructure.
• Choose the dashboard if you have a handful of files and no code to write.
• Check the privacy term against your data. Files auto-delete after 24 hours, which suits one-off jobs but may not suit workflows needing longer retention on the provider side.
• Confirm current pricing before committing. The site advertises pay-as-you-go with no hidden fees and a 5-credit free start, but exact per-credit rates and volume terms live on its pricing page — read them there rather than assuming.

Common sticking points

• Credits, not minutes, are the unit. The free tier is described in transcription credits, so check how credits map to file length before estimating cost.
• Model choice changes both speed and accuracy. large-v2 is the accuracy-oriented option shown in the docs; smaller models trade accuracy for turnaround.
• Language codes matter. Passing the wrong language value can degrade output even when the model supports the language.
• The API key is a secret. It travels in the X-API-Key header, so keep it server-side rather than in client code.