Of course this shows the exact same bad behavior as all current softorbits products in that it forces installation into an insecure subfolder of the logged in users appdata folder and downloads unattributed CC licensed AI model https://huggingface.co/istupakov/parakeet-ctc-0.6b-onnx
places it in a fixed location on the system drive also
C:\Users\[WINDOWSLOGINNAME]\AppData\Roaming\com.softorbits.voicedictation It also installs if not already present Microsoft Edgewebview2 NO GPU acceleration supported only multi-core CPU supported and while processing a wav file the computer becomes VERY sluggish or snail like. It is NOT implementing co-operative multi-tasking so to use the computer as intended the priority of the voicedictation.exe process needs to be lowered to below noraml, and then the windows UI wakes up again. This program is badly designed, badly programmed and badly installed. IMHO. The underlying voice to text transcribing model produced by hugging face user istupakov is quite good, shame about the GUI! It’s cooking my C:drive SSD at 54C! Awww the transcription of a 188MByte WAV file failed with a non-zero error value returned. There is also conflicting information with part of the software claiming it’s version 1.5 and other parts clearly state version 1.2 and update checking is disabled and while transcribing this wav file there is 45.5/51.1GB committed and 13-15.8GB of physical RAM in use. Add files defaults to just supported video file types and ignores audio file types… it should default to all supported file types! was unable to write the transcript txt file to the videos folder so 100% failure on all file transcriptions just works quite well on recorded speech via system microphone… but it is not real-time… it records the voice first then when recording is stopped the recording is processed. I’m deleting it all now.