Guides
Step-by-step guides from the first start to a two-way interpreted call. Each one says what it needs: a GPU, a virtual audio cable, or nothing special.
First start on a clean machine
From the cloned repository to the first caption – environment check, account service, keys and the daily launcher, in the real order, on Windows and macOS.
A Teams call with interpreting in both directions
Two virtual audio cables, the Usage tab's map, the Teams speaker and microphone, your own voice sample and the outgoing direction switch.
Captions and translation for lectures and video
The machine's audio mode on the Usage tab, the Accurate segmentation profile, the interpreter with or without headphones, and exporting recordings to SRT.
Subtitling a video folder to SRT
Starting the caption server, the scripts' key, the switches of transcribe-folder.ps1, the three views, the subtitle file's name and the exit codes.
Translating subtitle files, markdown and resource files
Three translation clients – subtitle, markdown, resource file – the key from an environment variable, the --check and --dry-run modes, and exit codes.
One GPU machine on the network, several workstations
The caption server, the translator and the voice service run on one GPU machine and workstations reach them over the network: firewall, TLS, keys.
Captions without a GPU using Windows Live Captions
Without a graphics card, captions can come from Windows' own speech recognition: what it can do, what it cannot, and how to switch it on.
Accounts, keys and permissions
The account service owns users, roles and API keys: first login, issuing keys, lost passwords, the audit log and usage metering.
Troubleshooting
Symptom first: no captions, missing translation, silent interpreter, unreachable server – the cause and what to do.