These models are for KoboldCpp's speech recognition. You would load them with `--whispermodel` or from the "Audio" tab in the GUI launcher.

![image/png](/static-proxy?url=https%3A%2F%2Fcdn-uploads.huggingface.co%2Fproduction%2Fuploads%2F63cd4b6d1c8a5d1d7d76a778%2FzPTgBos1StPjBHS51Rmj8.png)

Then in KoboldAI Lite, you can enable Voice Input in the settings (Push To Talk or Voice Detect)

![image/png](/static-proxy?url=https%3A%2F%2Fcdn-uploads.huggingface.co%2Fproduction%2Fuploads%2F63cd4b6d1c8a5d1d7d76a778%2FHuIbaoe5trKtnihCGpzaH.png)

NEW: Now we have Audio.cpp transcribe model support (ASR and forced aligner) added too, check the "audiocpp" folder