← All guides
Core Features

Live Audio

2026-10-27 · 3 min read
On this page

    What's it for

    Live Audio will listen to what your PC is playing when you want it to (streams, videos, anime, games, calls) and will show what is being said as subtitles, the subtitle will contain the Japanese that was said as well as the translation shown below it (the translation can take a small pause before appearing). It will work on anything that plays through your speakers or headphones, and does not use your microphone so it is specifically for playback.

    Before you start

    • The Live Audio add-on: Settings > Live Audio > Add-ons > Install ("Live Audio's speech models": faster-whisper, pyaudiowpatch, soxr; a minute or two). Live Audio won't start without it
    • A translator set up (pyontan AI works out of the box)
    • Optional: GPU acceleration (NVIDIA only, about 2 GB). Significant speedup (approx up to 10x faster), but CPU is fine for most people

    Steps

    Pick what it listens to

    • Settings > Live Audio > Audio source: the speakers or headphones the sound plays on. Refresh if a new device isn't listed.

    Start

    • Alt + T, Home Menu > Live Audio or tray menu > Live Audio Translation (would show a tick)
    • Subtitles appear at the bottom of the screen under it
    • A small Live mark appears in the corner while it runs, pulsing when detecting sound

    Stop

    • Alt + T again, or untick it from the tray menu (or press from home menu again)
    • Everything that was detected and said will be in the Voice History in History (Alt + H) > Voice

    Options Worth Knowing (Settings > Live Audio)

    • Mode: Segmented (waits for a pause based on pause length, more accurate) vs Streaming (faster, shows words as they come and recorrects what was heard)
    • Whisper model sizes: base, small (default), medium, large-v3. The bigger the model size the more accurate but slower it can be (system-dependent)
    • Pause length: how long a pause ends a line (mostly for segmented, will wait this long before translating what was said before the pause)
    • Subtitle monitor (changes where the subtitles show), position and size
    • Live indicator corner, or Off
    • Clips: keeps a short audio clip of recent lines (300 by default) so Anki cards can carry real audio and you can play audio clips back
    • Alt + C saves clips of the last 30 seconds
    • Training wheels (Settings > Learning): hide the translation until you press Alt + V
    • Daily time: Free tier is 6 hours, Premium is 12 hours, Super Premium is unlimited

    Common Issues

    • Nothing appears: wrong Audio source has been set, or nothing is playing on it. After a while it says 'Live Audio hasn't heard any speech yet" with a link to the setting
    • Music and loud backgrounds are generally harder to recognise, hence why there's Song Lyrics as a feature
    • A lag of a second or two is normal in Segmented mode, wait until the end of a phrase or there's adamant silence (also can depend on your system)
    • Falling behind on a slower PC: use a smaller model or try using Streaming mode
    Related guides
    Ready to try pyontan?
    Free covers the basics. No card needed.
    Download for Windows