Apr 11
slow transcription lagging behind competition
I have just downloaded a new free open-source voice-to-text app, TypeWhisper, and the speed in which it spits out the text in comparison to Paraspeech is frankly confusing. I have a lifetime subscription to Paraspeech, but found myself using it less and less due to the speed in which it transcribes, sometimes 10/15 seconds waiting. Is that a bug on my end? I switched to the smaller, less accurate models, but that really isn’t a fix or fair solution. I would rather not switch as I’ve built up a lot of history with Paraspeech, but the quality needs to improve for me to continue using it on a daily basis.
Reviewing
Thanks for reporting this. I moved it into Reviewing; if you can reproduce the 10-15 second delay on the current build, please export diagnostics right after a slow transcription and send them through support so we can attribute the delay to model, rewrite, insertion, or setup time.