I made an open-source Android transcription keyboard using Whisper AI. You can dictate with auto punctuation and translation to many languages. :)

[DevEmperor]

Dictate is an easy-to-use keyboard for transcribing and dictating. The app uses OpenAI Whisper in the background, which supports extremely accurate results for many different languages with punctuation and auto translation using GPT-4 Omni.

You can download the app from Google Play Store:
https://play.google.com/store/apps/details?id=net.devemperor.dictate

Here you can see it in action:
https://www.youtube.com/watch?v=PSvLRnHYleg

And this is the repository with the source code:
https://github.com/DevEmperor/Dictate

[ipstone]

nice interface! But my question is why sending the audio to openAI for the transcribing? why not just do on device transcription, such as this one:
https://play.google.com/store/apps/details?id=net.devemperor.dictate&pli=1
or this one:
https://play.google.com/store/apps/details?id=com.signalmaster.asr&hl=en_US

For the second one, the code is open at :
https://github.com/vilassn/whisper_android

If your keyboard is using these local transcription engine without sending audio out, it will be a really nice/popular app!

[KanyantaM]

Perhaps the accuracy of on device transcription isn't that good. It really depends on the quality of your device

[DevEmperor]

Hey, thanks for this great tip. However, I want to design the app for everyone and most smartphones are not that powerful to run a local Whisper model. Also, I want to achieve quick results and the API may cost some money and require an internet connection, but it is fast and reliable.
I will keep this feature in mind, but currently, I sadly don't have time to implement it.
Thanks anyway for your idea. :)

[fire17]

Hopefully both options will be available.. run a small bench to determine which will be default (if sample takes less than x time to transcribe use local by default, else use api)
Got here cause I'm looking for an open source keyboard with ai abilities.... Not just transcribing but also text completion (copilot style) and other features