Skip to content
repository radar
Voice, Vision & Multimodal

🎙️RealtimeSTT

ACTIVE STEADY

KoljaB/RealtimeSTT

A robust, efficient, low-latency speech-to-text library with advanced voice activity detection, wake word activation and instant transcription.

⭐ Very popular: 10k stars, gaining about 13 a week

View on GitHub ↗

repo profile

vintage 2023 3 years old
language Python
license MIT

momentum

total stars 10k
stars added last week +13
used by 275 repos & packages
commits / week 0 steady
issues closed 30% closed same-day, median
contributors 24
release cadence quarterly
last activity 2mo ago

durability

backing community / independent
openness permissive
bus factor 1 solo
top-author share 93% 6 mo

bus factor = how many people it takes to cover more than half the commits (6 months). 1 is a solo project; higher means the work is spread across a team. top-author share is the single busiest author's slice of those commits.

since we covered it

monthly average + 155/mo (+2%/mo) · + 2k total since PR#11

why it's a big deal

  • Gives developers a ready-made speech-to-text layer for voice assistants and live transcription, handling the awkward parts like knowing when someone starts and stops speaking.
  • Optional wake word activation via Porcupine or OpenWakeWord means an app can idle until called, useful for hands-free interfaces without keeping full transcription running.
  • Runs across Linux, macOS and Windows with microphone, file, stream and websocket input, so it fits both desktop apps and browser streaming setups.

under the hood

  • Python library defaulting to faster_whisper for transcription, with swappable engines including OpenAI Whisper, whisper.cpp, Moonshine, sherpa-onnx and Parakeet NeMo.
  • Voice activity detection uses WebRTC VAD and Silero VAD, and audio is handled as 16-bit mono PCM at 16 kHz with optional resampling.
  • Uses multiprocessing to isolate model inference and exposes event callbacks for recording, VAD, realtime text and wake word state, with optional CUDA and a FastAPI browser streaming example.

Radar summary, generated from the project's public sources

star history

PR#11 · 8k10k now Aug 2023Aug 2026
  1. PR#11 8k 2025-06-25
  2. now 10k + 2k since first covered

curve is sampled from GitHub's star history, plus our own daily readings since we covered it; the dashed stretch is before we first covered it, the solid line since. figures at coverage are the numbers we printed then (approx.), current count is live.

understory

Output and attention are roughly in balance.

-6 understory score output 42 · clout 48
Aug 2025 Jul 2026
  • output, commits & releases
  • clout, star velocity

output = commits & releases; clout = star velocity, both 0 to 100 monthly indices; the gap where output runs above clout is the understory. The understory →

covered in

  • PR#11 2025-06-25 below the radar

    Low-Latency Speech-to-Text with Wake Word Activation

similar projects

compare these →
  • 🎙️ dograh

    leaner, 5k stars

    Open source voice AI platform. Self-hosted alternative to Vapi and Retell. On Prem, BYOK across Speech to Speech or LLM/STT/TTS, with a visual workflow builder, MCP native and telephony support.

    5k ACTIVE
  • Unsloth

    7.1× the stars

    Local UI to run and train LLMs and diffusion models, including Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, DeepSeek-V4, FLUX and more.

    72k ACTIVE
  • 🎙️ Real-Time-Voice-Cloning

    6× the stars

    Clone a voice in 5 seconds to generate arbitrary speech in real-time

    60k ACTIVE

comments

Sign in with GitHub to add your blip on RealtimeSTT.

loading comments…