Open WesleyFister opened 6 months ago
It would be very nice to be able to record/speak a bit longer. Can a "stop recording" button be implemented? I can imagine the benefits of having a speaking limit but 30 seconds is a bit too little.
Would it be possible, for example, to allow a user to record/speak up to 5 minutes, and the user can stop whenever within this limit?
We already have a "stop recording" button. When recording the user can tap the microphone button to end the recording whenever they choose.
Thats true. :) Im a new user so I missed that "stop button". Still, longer recording time would be great. Is it feasible to prolong the speaking time or is there a technical reason behind 30 sec?
Whisper splits audio into 30 second chunks but that shouldn't be a limiting factor. Maybe, audio that is greater than 30 seconds experiences a massive slow down or something? I'm not entirely sure of the reason myself and would like to know this as well.
Thank you for the quick reply. : )
Found this post talking about the reason behind 30sec.
I also read this project where one can record a longer audio and then split them into 30sec chunks and merge the transcribed result (if im understanding it correct). But perhaps its not a neat solution for futo.
I really hope they can do an auto splitting thing so that we can indeed record longer than 30 seconds.
I would love to use the better speech recognition of whisper to do my multi minute long brain dumping sessions using speech
I really hope they can do an auto splitting thing so that we can indeed record longer than 30 seconds.
I would love to use the better speech recognition of whisper to do my multi minute long brain dumping sessions using speech
Not that I like the idea of brain dumps but this would be a great use case. In the meantime, if you're on Linux you can use SpeechNote that would do the same thing but with bigger/better models.
Not that I like the idea of brain dumps but this would be a great use case. In the meantime, if you're on Linux you can use SpeechNote that would do the same thing but with bigger/better models.
Thanks for the recommendation! (I'm on linux) so I'll check it out. Currently I'm just recording my voice and then putting it through Whishper https://github.com/pluja/whishper/
Allow the user to change or disable the recording limit and by default if the user is holding the microphone button disable the limit.