Vilas Ninawe
Vilas Ninawe
Please check your build environment. It can be due to ndk configuration or some other problem.
You need to implement VAD.
@sirramin I tried to run whisper tflite model in react native. But, no luck. If you are developing app for Android or iOS, you can run model in kotlin/java or...
@sirramin for android, you can write react native wrapper on top of java code. Are you able to use react native directly on iOS ?
Yes. We are planning to support live transcription from mic and long file with timestamps.
@michoael we are working on adding vad support, realtime transcription and noise processing. Supporting these features in Java will add significant delay. So using native code to add these feature...
Yes. We will support realtime transcription and long wav file transcription. It is under implementation.
Did you able to transcribe internal files? If yes, may be your app doesn't have permission to access external files. May be your file is not in 16k, mono, pcm_16/pcm_float...
audio classification is not same as whisper. It has different purpose like categorizing audio. While whisper is used for speech to text conversion.
Is your model loaded successfully in Android? Do you see any error or failure in logcat? Does your model works properly in python?