You need to agree to share your Hugging Face information to access this model
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
To get access, please fill in the short form at https://saidzoda.com/access. We approve requests after we receive the form.
Log in or Sign Up to review the conditions and access this model content.
Sado Transcriber 0.9.4 (Saidzoda Lab)
How to get access: please fill in the short form at saidzoda.com/access. We approve requests after we receive the form.
Self-hosted transcription of two-party call-center recordings in Tajik with Russian code-switching, with a browser review screen (timeline player, word-following transcript, inline correction), staff logins, HTTPS, PII tagging, and a training-dataset export. Runs on the customer's own server; audio never leaves it.
release/sado-transcriber-0.9.4.tar.gz: installer package (install.sh,run.sh,fetch_models.py, app code).- Models (gated, fetched by the installer):
Tohirju/sl-cyanct2-int8/custdomft-ckpt-1318(Large),Tohirju/sl-marjoramct2-int8/turbo-fullmix25(Turbo),Tohirju/sl-gentian/Tohirju/sl-lupin(punctuation / truecase, large-news),Tohirju/ecapa-tdnn-tajik(voiceprints). results/: tests on 20 recorded Sado calls whose callers are SIMULATED (Chatterbox TTS voices). These numbers are engineering checks only; they are not accuracy claims for real calls.
Tested 2026-10-16: clean install from an empty folder on macOS arm64 (Python 3.12) and on Ubuntu 24.04 (Docker, aarch64, Python 3.12.3); 26 unit tests; end-to-end login, upload, Large transcription, export, start/stop. Not yet tested: x86_64 Linux hardware, NVIDIA GPU path.