OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Audio & Voice AI

Whisper Playground

Whisper Playground is a self-hostable web application for real-time or sequential speech-to-text transcription with speaker diarization. Its Python backend uses faster-whisper for recognition, with Diart and Pyannote-based components for diarization, and its frontend is built with React.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Whisper Playground — OpenViST project profile in Audio & Voice AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want a self-hostable web application for speech transcription and need to choose between real-time output and sequential transcription with more context.

Who it is for

  • Developers building a web interface for speech transcription.
  • Teams or individuals who want to run a configurable transcription app on their own setup.

Documented uses

  • Transcribe speech in real time or in periodic sequential segments.
  • Pair transcription with speaker diarization.
  • Adjust the selected model size, language, timeout and beam size.
Before you choose

Requirements & limits

The repository is public and licensed under MIT. Running it locally requires Conda and Yarn. Pyannote diarization requires a Hugging Face account, acceptance of the relevant model terms and login with a user access token.

  • Real-time mode supports real-time diarization; sequential mode transcribes periodically with more context.
  • The timeout setting affects which audio is transcribed in real-time mode; the project notes that audio shorter than the configured timeout may not be transcribed.
  • The README reports possible uncontrolled speaker swapping in sequential mode.
  • The project says it has not been tested for all languages.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up and launch Whisper Playground locally

Install the backend and frontend environments, configure them, and start both parts of the application.

Before you start

  • Conda and Yarn
  • A clone or fork of the repository
  • A transcription device and compute type to set in config.py

Steps

  1. Clone or fork the repository.

  2. Run sh install_playground.sh to install the backend and frontend environments.

  3. Review config.py for the transcription device and compute type, and config.js for the backend configuration and address.

  4. Start the backend with cd backend && python server.py.

  5. In a separate terminal, start the React frontend with cd interface && yarn start.

Expected outcome

The Python backend and React frontend are started as separate processes.

Guide limitations

  • Speaker diarization additionally requires accepting the Pyannote model terms and logging in to Hugging Face with a user access token.
  • On macOS, if building the safetensors wheel fails, the README recommends installing Rust and trying again.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
saharmor/whisper-playground · Official homepage title · retained in its source language
How we review sourcesSuggest a correction