innovatorved/whisper.api

This project provides an API with user level access support to transcribe speech to text using a finetuned and processed Whisper ASR model.

What it solves

Whisper API provides a self-hosted, high-performance speech-to-text transcription service. It allows users to maintain full data ownership while using a standardized API that is compatible with Deepgram's /v1/listen endpoints, making it easy to integrate into existing applications without switching providers.

How it works

The project acts as an API wrapper around whisper.cpp, exposing the transcription capabilities of OpenAI's Whisper models via REST and WebSocket interfaces. It includes a CLI for managing API keys and models, and supports both file uploads and transcription from URLs with built-in SSRF protections.

Who it’s for

Developers and organizations that need a scalable, private transcription service that can be drop-in compatible with existing Deepgram-style workflows.

Highlights

  • Deepgram-compatible API: Drop-in compatibility with /v1/listen endpoints.
  • Real-time Streaming: Supports 16kHz PCM transcription via WebSockets.
  • ** Stabilization**: Native support for JSON, SRT, and VTT export formats.
  • Advanced Transcription: Includes speaker diarization, custom vocabulary prompting, and audio cropping.
  • Secure Management: CLI-based API key generation and secure model management.