An LLM-powered desktop and CLI tool for video subtitles: speech recognition, sentence segmentation, correction, translation and burning subtitles into video.
16.3K1.4KPythonGPL-3.0Updated last month
royshil/obs-localvocal
An OBS Studio plugin that transcribes speech to captions in real time and translates them, running Whisper models locally on your machine.
- Stars
- 1,642
- Forks
- 126
- Open issues
- 61
- Last commit
- 5 days ago
What is it?
LocalVocal is a plugin for OBS Studio that turns speech from a source into on-screen captions and text files. It runs OpenAI's Whisper models locally through whisper.cpp, so audio is processed on your own machine without cloud services, network access, or per-minute costs. Translation of the captions is handled with CTranslate2.
Captions can be shown through OBS text sources, written to .txt or .srt files, synced with recording timestamps, or sent over RTMP to platforms such as YouTube and Twitch. Users can pick from bundled and downloadable Whisper models, or load their own GGML model file. Acceleration is available through CUDA, hipBLAS for AMD ROCm, Vulkan, OpenCL, and Apple Metal or CoreML, with a CPU fallback.
Prebuilt installers are provided for Windows, Linux (Ubuntu .deb and Flatpak), and macOS, with separate generic, NVIDIA, and AMD builds on Windows and Linux.
Key features
- 01Real-time speech-to-text transcription in 100 languages using local Whisper models
- 02Translates captions into other languages, including via cloud providers or local NMT models
- 03Shows captions in OBS text sources and saves them to .txt or .srt files
- 04Sends captions to RTMP streams such as YouTube and Twitch
- 05Supports CUDA, AMD ROCm, Vulkan, OpenCL, and Apple Metal/CoreML acceleration
Use cases
- Adding live captions to a stream or recording in OBS without sending audio to the cloud
- Producing SRT subtitle files from recorded or live audio for video editing
- Translating a streamer's speech into another language for an international audience
- Running private transcription on a machine where audio must stay local
Quick start
- 1
Download the installer for your OS and hardware from the latest releases page
- 2
Run the downloaded installer or package to install the plugin
- 3
In the plugin settings, select a GPU acceleration backend if you want GPU use (CPU is the default)
Things to know
- Licensed under GPL-2.0, a copyleft license that limits how derivative code can be distributed.
- NVidia and AMD GPU variants need extra drivers or the CUDA toolkit, and Windows needs the latest MSVC runtime.
Topics
FAQ
- What is obs-localvocal?
- An OBS Studio plugin that transcribes speech to captions in real time and translates them, running Whisper models locally on your machine. LocalVocal is a plugin for OBS Studio that turns speech from a source into on-screen captions and text files. It runs OpenAI's Whisper models locally through whisper.cpp, so audio is processed on your own machine without cloud services, network access, or per-minute costs. Translation of the captions is handled with CTranslate2.
- Is obs-localvocal free to use? What license does it have?
- obs-localvocal is open source under the GPL-2.0 license. Check the license terms before using it in commercial products.
- Is obs-localvocal still maintained?
- Actively maintained. The last commit to obs-localvocal was on Oct 15, 2026.
- What can I use obs-localvocal for?
- Adding live captions to a stream or recording in OBS without sending audio to the cloud Producing SRT subtitle files from recorded or live audio for video editing Translating a streamer's speech into another language for an international audience Running private transcription on a machine where audio must stay local
- How do I install obs-localvocal?
- Download the installer for your OS and hardware from the latest releases page Run the downloaded installer or package to install the plugin In the plugin settings, select a GPU acceleration backend if you want GPU use (CPU is the default)
- Where can I get obs-localvocal?
- The source code and installation instructions are on GitHub: https://github.com/royshil/obs-localvocal
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
General-purpose speech recognition model that transcribes, translates and identifies the language of audio across many languages.
110.9K13.4KPythonMITUpdated 2 months ago
C/C++ port of OpenAI's Whisper speech recognition model for fast, offline, dependency-free speech-to-text on CPU and GPU.
54.5K6.3KC++MITUpdated yesterday
A faster reimplementation of OpenAI's Whisper speech-to-text model on the CTranslate2 inference engine, with lower memory use and 8-bit quantization.
26K2.2KPythonMITUpdated 4 days ago
Open-source speech-to-text built on Whisper that adds word-level timestamps via forced alignment and optional speaker diarization.
24.7K2.5KPythonBSD-2-ClauseUpdated 2 days ago
Self-hosted real-time speech-to-text server with streaming ASR, speaker diarization, live translation and OpenAI/Deepgram-compatible APIs.
11.2K1.1KPythonApache-2.0Updated 20 days ago