CPU-only CLI that transcribes any video/audio to sentence-level .srt
using whisper.cpp tiny.en.
Get the latest zip/tar from: https://github.com/appxa/tinystt/releases
| File | OS |
|---|---|
tinystt-linux-x64.tar.gz |
Linux |
tinystt-windows-x64.zip |
Windows |
tinystt-macos-x64.tar.gz |
macOS |
Each archive is only the binary. You still need ffmpeg and the model.
# 1) unpack binary onto PATH
tar -xzf tinystt-linux-x64.tar.gz # or tinystt-macos-x64.tar.gz
mkdir -p ~/.local/bin
mv tinystt ~/.local/bin/
chmod +x ~/.local/bin/tinystt
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc # mac: ~/.zshrc
source ~/.bashrc
# 2) ffmpeg
# Debian/Ubuntu: sudo apt install ffmpeg
# Fedora: sudo dnf install ffmpeg
# macOS: brew install ffmpeg
# 3) model (once; put next to video, or set TINYSTT_MODEL)
curl -L -o ggml-tiny.en.bin \
https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.en.bin
# 4) run
tinystt video.mp4 # → video.srt- Install ffmpeg and put it on PATH.
- Unzip
tinystt-windows-x64.zip→ you gettinystt.exe. - Put
tinystt.exesomewhere on PATH, or keep it next to your video. - Download the model once:
curl.exe -L -o ggml-tiny.en.bin https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-tiny.en.bin- Run (cmd / PowerShell):
tinystt.exe video.mp4
# or: .\tinystt.exe video.mp4Optional env vars (all OS):
TINYSTT_THREADS=2 tinystt video.mp4
TINYSTT_MODEL=/path/to/ggml-tiny.en.bin tinystt video.mp4Default: 1 CPU thread. Model from cwd or TINYSTT_MODEL.
Needs: C++17, cmake ≥ 3.22, g++/MSVC/clang, ffmpeg, git.
git clone https://github.com/appxa/tinystt.git
cd tinystt
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j$(nproc)
install -m 755 build/tinystt ~/.local/bin/tinystt # Windows: copy build\Release\tinystt.exewhisper.cpp is fetched at build time via CMake FetchContent — no submodules.
Maintainers: tag v0.1.0 (or any v*) and push to publish release zips via CI.
MIT