2 ms·
I've been using their Voice Input for a few months now and really like it. My only issue is that it buffers the input and then spews it all when you stop speaki
by skeledrew 2y ago
I've been using their Voice Input for a few months now and really like it. My only issue is that it buffers the input and then spews it all when you stop speaking, but I guess that helps the model to be more contextually accurate.
I hope their Keyboard is as good and provides decent multilingual support. Currently using Heliboard, and have to switch to a broken GO Keyboard install (dictionary is missing), that failed to properly clone from my previous device, whenever I want to type Japanese.
- Ey7NFZ3P0nzAe 2y agoThe fact that you have to wait the end of the 30 seconds before it starts writing the text is a fundamental limitation of the current transformer algorithm because the first and the last bits of speech influence each other so you have to wait until the end.
- skeledrew 2y agoYeah I get that. I tried another open source voice input that was real time and the quality was horrible. But this is something that can be worked around for Whisper. One thing that comes to mind is an option to append and reprocess the audio every few centiseconds (needs a fairly powerful device though), and update the text output as needed. This could also open the door for an edit-by-voice feature.