AI Transcript

The AI Transcript App is a full-stack tool that records your voice, transcribes it locally using Whisper, and cleans the resulting text with an LLM — removing filler words, fixing grammar, and preserving meaning.

It’s designed to run entirely on your machine with no cloud dependency, though it supports any OpenAI-compatible cloud API as a drop-in replacement.

Features:

  • Browser-based voice recording (hold V to record)
  • Audio file upload via drag-and-drop
  • Whisper speech-to-text (runs locally via faster-whisper)
  • LLM-powered transcript cleaning (removes filler words, fixes errors)
  • Manual text input — paste any text for LLM cleaning
  • Customizable LLM system prompt
  • One-click copy to clipboard
  • OpenAI API-compatible — works with Ollama, LM Studio, OpenAI, or any compatible provider

Stack:

LayerTechnology
FrontendReact 19, TypeScript, Vite
BackendFastAPI, Python 3.12, uv
Speech-to-Textfaster-whisper (local)
LLMOpenAI Python SDK → Ollama (default)
ContainerizationDocker Compose

Project link: https://github.com/ruchirV/local-ai-transcript-app