Download Info
- Author: k3ct
- File size: 17.19 kB
- Downloads: 313
- Updated: 2025-10-18
- Categories:
If the documentation changes, the date in the filename will change.
October 18 - TTS_WSL2_model_gen.txt file change only.
Added additional instructions for GPU training and model export.
October 17 - TTS_WSL2_model_gen.txt file change only.
Added this text:
Modern soundcards are configured for an electret MIC and provide the necessary bias. A Heil headset with a dynamic MIC will not generate enough recording audio for your recordings.
If you have more than one soundcard in your PC, the recording script uses the Default MIC for recording. If you are unable to change Windows so your recording MIC input is the Default device, there is a version of the recording script that hasn't been tested which will allow you to list the sound devices and select the device that you want to use. Contact K3CT if you need this script version.
If you bypass the recording script and use a different recording program, it will bypass the dynamic range checks and downsampling built into the recording script. This could lead to a poor quality model or if the wav file format is incorrect a run-time error during training.
